<target name="copy_resources" description="Copies resources">
<mkdir dir="${classes.dir}/resources" />
<copy verbose="true"
- file="../resources/phyloxml_schema/1.10/phyloxml.xsd"
+ file="../resources/phyloxml_schema/1.20/phyloxml.xsd"
todir="${classes.dir}/resources"/>
</target>
-<?xml version="1.0" encoding="UTF-8"?>\r
-<!-- -->\r
-<!-- phyloXML -->\r
-<!-- -->\r
-<!-- schema in XMLSchema -->\r
-<!-- -->\r
-<!-- License: The phyloXML XML Schema Definition is -->\r
-<!-- dual-licensed under the LGPL or Ruby's -->\r
-<!-- License. -->\r
-<!-- You can redistribute and/or modify -->\r
-<!-- it under either the terms of the LGPL, -->\r
-<!-- or Ruby's License. -->\r
-<!-- see: http://www.ruby-lang.org/en/about/license.txt -->\r
-<!-- -->\r
-<!-- Copyright (c) 2008-2013 Christian Zmasek -->\r
-<!-- -->\r
-<!-- www.phyloxml.org -->\r
-<!-- Version: 1.16 -->\r
-<!-- Last modified: 2013.09.30 by Christian Zmasek -->\r
-<!-- -->\r
-<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:phy="http://www.phyloxml.org"\r
- targetNamespace="http://www.phyloxml.org" elementFormDefault="qualified" attributeFormDefault="unqualified">\r
- <xs:annotation>\r
- <xs:documentation> phyloXML is an XML language to describe evolutionary trees and associated data. Version: 1.10.\r
- License: dual-licensed under the LGPL or Ruby's License. Copyright (c) 2008-2011 Christian M Zmasek.</xs:documentation>\r
- </xs:annotation>\r
- <!-- phyloxml is the root element:-->\r
- <xs:element name="phyloxml" type="phy:Phyloxml"/>\r
- <!-- phyloXML definition:-->\r
- <xs:complexType name="Phyloxml">\r
- <xs:annotation>\r
- <xs:documentation> 'phyloxml' is the name of the root element. Phyloxml contains an arbitrary number of\r
- 'phylogeny' elements (each representing one phylogeny) possibly followed by elements from other namespaces.\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence maxOccurs="unbounded">\r
- <xs:element name="phylogeny" type="phy:Phylogeny" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- Phylogeny:-->\r
- <xs:complexType name="Phylogeny">\r
- <xs:annotation>\r
- <xs:documentation> Element Phylogeny is used to represent a phylogeny. The required attribute 'rooted' is used\r
- to indicate whether the phylogeny is rooted or not. The attribute 'rerootable' can be used to indicate that\r
- the phylogeny is not allowed to be rooted differently (i.e. because it is associated with root dependent\r
- data, such as gene duplications). The attribute 'type' can be used to indicate the type of phylogeny (i.e.\r
- 'gene tree'). It is recommended to use the attribute 'branch_length_unit' if the phylogeny has branch\r
- lengths. Element clade is used in a recursive manner to describe the topology of a phylogenetic\r
- tree.</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="id" type="phy:Id" minOccurs="0"/>\r
- <xs:element name="description" type="xs:token" minOccurs="0"/>\r
- <xs:element name="date" type="xs:dateTime" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="clade" type="phy:Clade" minOccurs="0"/>\r
- <xs:element name="clade_relation" type="phy:CladeRelation" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="sequence_relation" type="phy:SequenceRelation" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="rooted" type="xs:boolean" use="required"/>\r
- <xs:attribute name="rerootable" type="xs:boolean"/>\r
- <xs:attribute name="branch_length_unit" type="xs:token"/>\r
- <xs:attribute name="type" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- Clade:-->\r
- <xs:complexType name="Clade">\r
- <xs:annotation>\r
- <xs:documentation> Element Clade is used in a recursive manner to describe the topology of a phylogenetic tree.\r
- The parent branch length of a clade can be described either with the 'branch_length' element or the\r
- 'branch_length' attribute (it is not recommended to use both at the same time, though). Usage of the\r
- 'branch_length' attribute allows for a less verbose description. Element 'confidence' is used to indicate\r
- the support for a clade/parent branch. Element 'events' is used to describe such events as gene-duplications\r
- at the root node/parent branch of a clade. Element 'width' is the branch width for this clade (including\r
- parent branch). Both 'color' and 'width' elements apply for the whole clade unless overwritten in-sub\r
- clades. Attribute 'id_source' is used to link other elements to a clade (on the xml-level).\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="branch_length" type="xs:double" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="width" type="xs:double" minOccurs="0"/>\r
- <xs:element name="color" type="phy:BranchColor" minOccurs="0"/>\r
- <xs:element name="taxonomy" type="phy:Taxonomy" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="sequence" type="phy:Sequence" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="events" type="phy:Events" minOccurs="0"/>\r
- <xs:element name="binary_characters" type="phy:BinaryCharacters" minOccurs="0"/>\r
- <xs:element name="distribution" type="phy:Distribution" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="date" type="phy:Date" minOccurs="0"/>\r
- <xs:element name="reference" type="phy:Reference" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="clade" type="phy:Clade" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="branch_length" type="xs:double"/>\r
- <xs:attribute name="id_source" type="phy:id_source"/>\r
- <xs:attribute name="collapse" type="xs:boolean"/>\r
- </xs:complexType>\r
- <!-- Taxonomy:-->\r
- <xs:complexType name="Taxonomy">\r
- <xs:annotation>\r
- <xs:documentation> Element Taxonomy is used to describe taxonomic information for a clade. Element 'code' is\r
- intended to store UniProt/Swiss-Prot style organism codes (e.g. 'APLCA' for the California sea hare 'Aplysia\r
- californica') or other styles of mnemonics (e.g. 'Aca'). Element 'authority' is used to keep the authority,\r
- such as 'J. G. Cooper, 1863', associated with the 'scientific_name'. Element 'id' is used for a unique\r
- identifier of a taxon (for example '6500' with 'ncbi_taxonomy' as 'provider' for the California sea hare).\r
- Attribute 'id_source' is used to link other elements to a taxonomy (on the xml-level).</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="id" type="phy:Id" minOccurs="0"/>\r
- <xs:element name="code" type="phy:TaxonomyCode" minOccurs="0"/>\r
- <xs:element name="scientific_name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="authority" type="xs:token" minOccurs="0"/>\r
- <xs:element name="common_name" type="xs:token" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="synonym" type="xs:token" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="rank" type="phy:Rank" minOccurs="0"/>\r
- <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="id_source" type="phy:id_source"/>\r
- </xs:complexType>\r
- <xs:simpleType name="TaxonomyCode">\r
- <xs:restriction base="xs:token">\r
- <xs:pattern value="([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:simpleType name="Rank">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="domain"/>\r
- <xs:enumeration value="superkingdom"/>\r
- <xs:enumeration value="kingdom"/>\r
- <xs:enumeration value="subkingdom"/>\r
- <xs:enumeration value="branch"/>\r
- <xs:enumeration value="infrakingdom"/>\r
- <xs:enumeration value="superphylum"/>\r
- <xs:enumeration value="phylum"/>\r
- <xs:enumeration value="subphylum"/>\r
- <xs:enumeration value="infraphylum"/>\r
- <xs:enumeration value="microphylum"/>\r
- <xs:enumeration value="superdivision"/>\r
- <xs:enumeration value="division"/>\r
- <xs:enumeration value="subdivision"/>\r
- <xs:enumeration value="infradivision"/>\r
- <xs:enumeration value="superclass"/>\r
- <xs:enumeration value="class"/>\r
- <xs:enumeration value="subclass"/>\r
- <xs:enumeration value="infraclass"/>\r
- <xs:enumeration value="superlegion"/>\r
- <xs:enumeration value="legion"/>\r
- <xs:enumeration value="sublegion"/>\r
- <xs:enumeration value="infralegion"/>\r
- <xs:enumeration value="supercohort"/>\r
- <xs:enumeration value="cohort"/>\r
- <xs:enumeration value="subcohort"/>\r
- <xs:enumeration value="infracohort"/>\r
- <xs:enumeration value="magnorder"/>\r
- <xs:enumeration value="superorder"/>\r
- <xs:enumeration value="order"/>\r
- <xs:enumeration value="suborder"/>\r
- <xs:enumeration value="infraorder"/>\r
- <xs:enumeration value="superfamily"/>\r
- <xs:enumeration value="family"/>\r
- <xs:enumeration value="subfamily"/>\r
- <xs:enumeration value="supertribe"/>\r
- <xs:enumeration value="tribe"/>\r
- <xs:enumeration value="subtribe"/>\r
- <xs:enumeration value="infratribe"/>\r
- <xs:enumeration value="genus"/>\r
- <xs:enumeration value="subgenus"/>\r
- <xs:enumeration value="superspecies"/>\r
- <xs:enumeration value="species"/>\r
- <xs:enumeration value="subspecies"/>\r
- <xs:enumeration value="variety"/>\r
- <xs:enumeration value="varietas"/>\r
- <xs:enumeration value="subvariety"/>\r
- <xs:enumeration value="form"/>\r
- <xs:enumeration value="subform"/>\r
- <xs:enumeration value="cultivar"/>\r
- <xs:enumeration value="strain"/>\r
- <xs:enumeration value="section"/>\r
- <xs:enumeration value="subsection"/>\r
- <xs:enumeration value="unknown"/>\r
- <xs:enumeration value="other"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!-- Sequence:-->\r
- <xs:complexType name="Sequence">\r
- <xs:annotation>\r
- <xs:documentation> Element Sequence is used to represent a molecular sequence (Protein, DNA, RNA) associated\r
- with a node. 'symbol' is a short (maximal 20 characters) symbol of the sequence (e.g. 'ACTM') whereas\r
- 'name' is used for the full name (e.g. 'muscle Actin'). 'gene_name' can be used when protein and gene names differ.\r
- 'location' is used for the location of a sequence on a genome/chromosome. The actual sequence can be stored with the \r
- 'mol_seq' element. Attribute 'type' is used to indicate the type of sequence ('dna', 'rna', or 'protein'). \r
- One intended use for 'id_ref' is to link a sequence to a taxonomy (via the taxonomy's 'id_source') in case \r
- of multiple sequences and taxonomies per node. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="symbol" type="phy:SequenceSymbol" minOccurs="0"/>\r
- <xs:element name="accession" type="phy:Accession" minOccurs="0"/>\r
- <xs:element name="name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="gene_name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="location" type="xs:token" minOccurs="0"/>\r
- <xs:element name="mol_seq" type="phy:MolSeq" minOccurs="0"/>\r
- <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="annotation" type="phy:Annotation" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="cross_references" type="phy:CrossReferences" minOccurs="0"/>\r
- <xs:element name="domain_architecture" type="phy:DomainArchitecture" minOccurs="0"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="type" type="phy:SequenceType"/>\r
- <xs:attribute name="id_source" type="phy:id_source"/>\r
- <xs:attribute name="id_ref" type="phy:id_ref"/>\r
- </xs:complexType>\r
- <xs:simpleType name="SequenceSymbol">\r
- <xs:restriction base="xs:token">\r
- <xs:pattern value="\S{1,20}"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:complexType name="MolSeq">\r
- <xs:annotation>\r
- <xs:documentation> Element 'mol_seq' is used to store molecular sequences. The 'is_aligned' attribute is used\r
- to indicated that this molecular sequence is aligned with all other sequences in the same phylogeny for\r
- which 'is aligned' is true as well (which, in most cases, means that gaps were introduced, and that all\r
- sequences for which 'is aligned' is true must have the same length).</xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="is_aligned" type="xs:boolean"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <xs:simpleType name="SequenceType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="rna"/>\r
- <xs:enumeration value="dna"/>\r
- <xs:enumeration value="protein"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!-- Accession:-->\r
- <xs:complexType name="Accession">\r
- <xs:annotation>\r
- <xs:documentation> Element Accession is used to capture the local part in a sequence identifier (e.g. 'P17304'\r
- in 'UniProtKB:P17304', in which case the 'source' attribute would be 'UniProtKB'). </xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="source" type="xs:token" use="required"/>\r
- <xs:attribute name="comment" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- CrossReferences: -->\r
- <xs:complexType name="CrossReferences">\r
- <xs:annotation>\r
- <xs:documentation> Used to store accessions to additional resources. </xs:documentation>\r
- </xs:annotation> \r
- <xs:sequence>\r
- <xs:element name="accession" type="phy:Accession" minOccurs="1" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType> \r
- <!-- DomainArchitecture: -->\r
- <xs:complexType name="DomainArchitecture">\r
- <xs:annotation>\r
- <xs:documentation> This is used describe the domain architecture of a protein. Attribute 'length' is the total\r
- length of the protein</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="domain" type="phy:ProteinDomain" minOccurs="1" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- <xs:attribute name="length" type="xs:nonNegativeInteger"/>\r
- </xs:complexType>\r
- <xs:complexType name="ProteinDomain">\r
- <xs:annotation>\r
- <xs:documentation> To represent an individual domain in a domain architecture. The name/unique identifier is\r
- described via the 'id' attribute. 'confidence' can be used to store (i.e.) E-values.</xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="from" type="xs:nonNegativeInteger" use="required"/>\r
- <xs:attribute name="to" type="xs:nonNegativeInteger" use="required"/>\r
- <xs:attribute name="confidence" type="xs:double"/>\r
- <xs:attribute name="id" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Events:-->\r
- <xs:complexType name="Events">\r
- <xs:annotation>\r
- <xs:documentation> Events at the root node of a clade (e.g. one gene duplication). </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="type" type="phy:EventType" minOccurs="0"/>\r
- <xs:element name="duplications" type="xs:nonNegativeInteger" minOccurs="0"/>\r
- <xs:element name="speciations" type="xs:nonNegativeInteger" minOccurs="0"/>\r
- <xs:element name="losses" type="xs:nonNegativeInteger" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <xs:simpleType name="EventType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="transfer"/>\r
- <xs:enumeration value="fusion"/>\r
- <xs:enumeration value="speciation_or_duplication"/>\r
- <xs:enumeration value="other"/>\r
- <xs:enumeration value="mixed"/>\r
- <xs:enumeration value="unassigned"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!--BinaryCharacters:-->\r
- <xs:complexType name="BinaryCharacters">\r
- <xs:annotation>\r
- <xs:documentation> The names and/or counts of binary characters present, gained, and lost at the root of a\r
- clade. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="gained" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- <xs:element name="lost" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- <xs:element name="present" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- <xs:element name="absent" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="type" type="xs:token"/>\r
- <xs:attribute name="gained_count" type="xs:nonNegativeInteger"/>\r
- <xs:attribute name="lost_count" type="xs:nonNegativeInteger"/>\r
- <xs:attribute name="present_count" type="xs:nonNegativeInteger"/>\r
- <xs:attribute name="absent_count" type="xs:nonNegativeInteger"/>\r
- </xs:complexType>\r
- <xs:complexType name="BinaryCharacterList">\r
- <xs:sequence>\r
- <xs:element name="bc" type="xs:token" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- Reference:-->\r
- <xs:complexType name="Reference">\r
- <xs:annotation>\r
- <xs:documentation> A literature reference for a clade. It is recommended to use the 'doi' attribute instead of\r
- the free text 'desc' element whenever possible. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="doi" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- Annotation:-->\r
- <xs:complexType name="Annotation">\r
- <xs:annotation>\r
- <xs:documentation> The annotation of a molecular sequence. It is recommended to annotate by using the optional\r
- 'ref' attribute (some examples of acceptable values for the ref attribute: 'GO:0008270',\r
- 'KEGG:Tetrachloroethene degradation', 'EC:1.1.1.1'). Optional element 'desc' allows for a free text\r
- description. Optional element 'confidence' is used to state the type and value of support for a annotation.\r
- Similarly, optional attribute 'evidence' is used to describe the evidence for a annotation as free text\r
- (e.g. 'experimental'). Optional element 'property' allows for further, typed and referenced annotations from\r
- external resources.</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- <xs:attribute name="ref" type="phy:ref"/>\r
- <xs:attribute name="source" type="xs:token"/>\r
- <xs:attribute name="evidence" type="xs:token"/>\r
- <xs:attribute name="type" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- Property:-->\r
- <xs:complexType name="Property" mixed="true">\r
- <xs:annotation>\r
- <xs:documentation> Property allows for typed and referenced properties from external resources to be attached\r
- to 'Phylogeny', 'Clade', and 'Annotation'. The value of a property is its mixed (free text) content.\r
- Attribute 'datatype' indicates the type of a property and is limited to xsd-datatypes (e.g. 'xsd:string',\r
- 'xsd:boolean', 'xsd:integer', 'xsd:decimal', 'xsd:float', 'xsd:double', 'xsd:date', 'xsd:anyURI'). Attribute\r
- 'applies_to' indicates the item to which a property applies to (e.g. 'node' for the parent node of a clade,\r
- 'parent_branch' for the parent branch of a clade). Attribute 'id_ref' allows to attached a property\r
- specifically to one element (on the xml-level). Optional attribute 'unit' is used to indicate the unit of\r
- the property. An example: <property datatype="xsd:integer" ref="NOAA:depth" applies_to="clade"\r
- unit="METRIC:m"> 200 </property> </xs:documentation>\r
- </xs:annotation>\r
- <xs:attribute name="ref" type="phy:ref" use="required"/>\r
- <xs:attribute name="unit" type="phy:ref"/>\r
- <xs:attribute name="datatype" type="phy:PropertyDataType" use="required"/>\r
- <xs:attribute name="applies_to" type="phy:AppliesTo" use="required"/>\r
- <xs:attribute name="id_ref" type="phy:id_ref"/>\r
- </xs:complexType>\r
- <xs:simpleType name="ref">\r
- <xs:restriction base="xs:token">\r
- <xs:pattern value="[a-zA-Z0-9_]+:\S+"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:simpleType name="AppliesTo">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="phylogeny"/>\r
- <xs:enumeration value="clade"/>\r
- <xs:enumeration value="node"/>\r
- <xs:enumeration value="annotation"/>\r
- <xs:enumeration value="parent_branch"/>\r
- <xs:enumeration value="other"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:simpleType name="PropertyDataType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="xsd:string"/>\r
- <xs:enumeration value="xsd:boolean"/>\r
- <xs:enumeration value="xsd:decimal"/>\r
- <xs:enumeration value="xsd:float"/>\r
- <xs:enumeration value="xsd:double"/>\r
- <xs:enumeration value="xsd:duration"/>\r
- <xs:enumeration value="xsd:dateTime"/>\r
- <xs:enumeration value="xsd:time"/>\r
- <xs:enumeration value="xsd:date"/>\r
- <xs:enumeration value="xsd:gYearMonth"/>\r
- <xs:enumeration value="xsd:gYear"/>\r
- <xs:enumeration value="xsd:gMonthDay"/>\r
- <xs:enumeration value="xsd:gDay"/>\r
- <xs:enumeration value="xsd:gMonth"/>\r
- <xs:enumeration value="xsd:hexBinary"/>\r
- <xs:enumeration value="xsd:base64Binary"/>\r
- <xs:enumeration value="xsd:anyURI"/>\r
- <xs:enumeration value="xsd:normalizedString"/>\r
- <xs:enumeration value="xsd:token"/>\r
- <xs:enumeration value="xsd:integer"/>\r
- <xs:enumeration value="xsd:nonPositiveInteger"/>\r
- <xs:enumeration value="xsd:negativeInteger"/>\r
- <xs:enumeration value="xsd:long"/>\r
- <xs:enumeration value="xsd:int"/>\r
- <xs:enumeration value="xsd:short"/>\r
- <xs:enumeration value="xsd:byte"/>\r
- <xs:enumeration value="xsd:nonNegativeInteger"/>\r
- <xs:enumeration value="xsd:unsignedLong"/>\r
- <xs:enumeration value="xsd:unsignedInt"/>\r
- <xs:enumeration value="xsd:unsignedShort"/>\r
- <xs:enumeration value="xsd:unsignedByte"/>\r
- <xs:enumeration value="xsd:positiveInteger"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!--Uri-->\r
- <xs:complexType name="Uri">\r
- <xs:annotation>\r
- <xs:documentation> A uniform resource identifier. In general, this is expected to be an URL (for example, to\r
- link to an image on a website, in which case the 'type' attribute might be 'image' and 'desc' might be\r
- 'image of a California sea hare'). </xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:anyURI">\r
- <xs:attribute name="desc" type="xs:token"/>\r
- <xs:attribute name="type" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Confidence:-->\r
- <xs:complexType name="Confidence">\r
- <xs:annotation>\r
- <xs:documentation> A general purpose confidence element. For example this can be used to express the bootstrap\r
- support value of a clade (in which case the 'type' attribute is 'bootstrap').</xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:double">\r
- <xs:attribute name="type" type="xs:token" use="required"/>\r
- <xs:attribute name="stddev" type="xs:double"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Identifier:-->\r
- <xs:complexType name="Id">\r
- <xs:annotation>\r
- <xs:documentation> A general purpose identifier element. Allows to indicate the provider (or authority) of an\r
- identifier. </xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="provider" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Distribution:-->\r
- <xs:complexType name="Distribution">\r
- <xs:annotation>\r
- <xs:documentation> The geographic distribution of the items of a clade (species, sequences), intended for\r
- phylogeographic applications. The location can be described either by free text in the 'desc' element and/or\r
- by the coordinates of one or more 'Points' (similar to the 'Point' element in Google's KML format) or by\r
- 'Polygons'. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- <xs:element name="point" type="phy:Point" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="polygon" type="phy:Polygon" minOccurs="0" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <xs:complexType name="Point">\r
- <xs:annotation>\r
- <xs:documentation> The coordinates of a point with an optional altitude (used by element 'Distribution').\r
- Required attributes are the 'geodetic_datum' used to indicate the geodetic datum (also called 'map datum',\r
- for example Google's KML uses 'WGS84'). Attribute 'alt_unit' is the unit for the altitude (e.g. 'meter').\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="lat" type="xs:decimal"/>\r
- <xs:element name="long" type="xs:decimal"/>\r
- <xs:element name="alt" type="xs:decimal" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="geodetic_datum" type="xs:token" use="required"/>\r
- <xs:attribute name="alt_unit" type="xs:token"/>\r
- </xs:complexType>\r
- <xs:complexType name="Polygon">\r
- <xs:annotation>\r
- <xs:documentation> A polygon defined by a list of 'Points' (used by element 'Distribution').\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="point" type="phy:Point" minOccurs="3" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- Date:-->\r
- <xs:complexType name="Date">\r
- <xs:annotation>\r
- <xs:documentation> A date associated with a clade/node. Its value can be numerical by using the 'value' element\r
- and/or free text with the 'desc' element' (e.g. 'Silurian'). If a numerical value is used, it is recommended\r
- to employ the 'unit' attribute to indicate the type of the numerical value (e.g. 'mya' for 'million years\r
- ago'). The elements 'minimum' and 'maximum' are used the indicate a range/confidence\r
- interval</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- <xs:element name="value" type="xs:decimal" minOccurs="0"/>\r
- <xs:element name="minimum" type="xs:decimal" minOccurs="0"/>\r
- <xs:element name="maximum" type="xs:decimal" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="unit" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- BranchColor:-->\r
- <xs:complexType name="BranchColor">\r
- <xs:annotation>\r
- <xs:documentation> This indicates the color of a clade when rendered (the color applies to the whole clade\r
- unless overwritten by the color(s) of sub clades).</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="red" type="xs:unsignedByte"/>\r
- <xs:element name="green" type="xs:unsignedByte"/>\r
- <xs:element name="blue" type="xs:unsignedByte"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- SequenceRelation:-->\r
- <xs:complexType name="SequenceRelation">\r
- <xs:annotation>\r
- <xs:documentation> This is used to express a typed relationship between two sequences. For example it could be\r
- used to describe an orthology (in which case attribute 'type' is 'orthology'). </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="distance" type="xs:double"/>\r
- <xs:attribute name="type" type="phy:SequenceRelationType" use="required"/>\r
- </xs:complexType>\r
- <xs:simpleType name="SequenceRelationType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="orthology"/>\r
- <xs:enumeration value="one_to_one_orthology"/>\r
- <xs:enumeration value="super_orthology"/>\r
- <xs:enumeration value="paralogy"/>\r
- <xs:enumeration value="ultra_paralogy"/>\r
- <xs:enumeration value="xenology"/>\r
- <xs:enumeration value="unknown"/>\r
- <xs:enumeration value="other"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!-- CladeRelation:-->\r
- <xs:complexType name="CladeRelation">\r
- <xs:annotation>\r
- <xs:documentation> This is used to express a typed relationship between two clades. For example it could be\r
- used to describe multiple parents of a clade.</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="distance" type="xs:double"/>\r
- <xs:attribute name="type" type="xs:token" use="required"/>\r
- </xs:complexType>\r
- <!-- Used to link elements together on the xml level:-->\r
- <xs:simpleType name="id_source">\r
- <xs:restriction base="xs:ID"/>\r
- </xs:simpleType>\r
- <xs:simpleType name="id_ref">\r
- <xs:restriction base="xs:IDREF"/>\r
- </xs:simpleType>\r
-</xs:schema>\r
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- -->
+<!-- phyloXML -->
+<!-- -->
+<!-- schema in XMLSchema -->
+<!-- -->
+<!-- License: The phyloXML XML Schema Definition is -->
+<!-- dual-licensed under the LGPL or Ruby's -->
+<!-- License. -->
+<!-- You can redistribute and/or modify -->
+<!-- it under either the terms of the LGPL, -->
+<!-- or Ruby's License. -->
+<!-- see: http://www.ruby-lang.org/en/about/license.txt -->
+<!-- -->
+<!-- Copyright (c) 2008-2016 Christian Zmasek -->
+<!-- -->
+<!-- www.phyloxml.org -->
+<!-- Version: 1.20 -->
+<!-- Last modified: 2016.07.06 by Christian Zmasek -->
+<!-- -->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:phy="http://www.phyloxml.org"
+ targetNamespace="http://www.phyloxml.org" elementFormDefault="qualified" attributeFormDefault="unqualified">
+ <xs:annotation>
+ <xs:documentation> phyloXML is an XML language to describe evolutionary trees and associated data. Version: 1.10.
+ License: dual-licensed under the LGPL or Ruby's License. Copyright (c) 2008-2011 Christian M Zmasek.</xs:documentation>
+ </xs:annotation>
+ <!-- phyloxml is the root element:-->
+ <xs:element name="phyloxml" type="phy:Phyloxml"/>
+ <!-- phyloXML definition:-->
+ <xs:complexType name="Phyloxml">
+ <xs:annotation>
+ <xs:documentation> 'phyloxml' is the name of the root element. Phyloxml contains an arbitrary number of
+ 'phylogeny' elements (each representing one phylogeny) possibly followed by elements from other namespaces.
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence maxOccurs="unbounded">
+ <xs:element name="phylogeny" type="phy:Phylogeny" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- Phylogeny:-->
+ <xs:complexType name="Phylogeny">
+ <xs:annotation>
+ <xs:documentation> Element Phylogeny is used to represent a phylogeny. The required attribute 'rooted' is used
+ to indicate whether the phylogeny is rooted or not. The attribute 'rerootable' can be used to indicate that
+ the phylogeny is not allowed to be rooted differently (i.e. because it is associated with root dependent
+ data, such as gene duplications). The attribute 'type' can be used to indicate the type of phylogeny (i.e.
+ 'gene tree'). It is recommended to use the attribute 'branch_length_unit' if the phylogeny has branch
+ lengths. Element clade is used in a recursive manner to describe the topology of a phylogenetic
+ tree.</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="name" type="xs:token" minOccurs="0"/>
+ <xs:element name="id" type="phy:Id" minOccurs="0"/>
+ <xs:element name="description" type="xs:token" minOccurs="0"/>
+ <xs:element name="date" type="xs:dateTime" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="clade" type="phy:Clade" minOccurs="0"/>
+ <xs:element name="clade_relation" type="phy:CladeRelation" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="sequence_relation" type="phy:SequenceRelation" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="rooted" type="xs:boolean" use="required"/>
+ <xs:attribute name="rerootable" type="xs:boolean"/>
+ <xs:attribute name="branch_length_unit" type="xs:token"/>
+ <xs:attribute name="type" type="xs:token"/>
+ </xs:complexType>
+ <!-- Clade:-->
+ <xs:complexType name="Clade">
+ <xs:annotation>
+ <xs:documentation> Element Clade is used in a recursive manner to describe the topology of a phylogenetic tree.
+ The parent branch length of a clade can be described either with the 'branch_length' element or the
+ 'branch_length' attribute (it is not recommended to use both at the same time, though). Usage of the
+ 'branch_length' attribute allows for a less verbose description. Element 'confidence' is used to indicate
+ the support for a clade/parent branch. Element 'events' is used to describe such events as gene-duplications
+ at the root node/parent branch of a clade. Element 'width' is the branch width for this clade (including
+ parent branch). Both 'color' and 'width' elements apply for the whole clade unless overwritten in-sub
+ clades. Attribute 'id_source' is used to link other elements to a clade (on the xml-level).
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="name" type="xs:token" minOccurs="0"/>
+ <xs:element name="branch_length" type="xs:double" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="width" type="xs:double" minOccurs="0"/>
+ <xs:element name="color" type="phy:BranchColor" minOccurs="0"/>
+ <xs:element name="taxonomy" type="phy:Taxonomy" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="sequence" type="phy:Sequence" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="events" type="phy:Events" minOccurs="0"/>
+ <xs:element name="binary_characters" type="phy:BinaryCharacters" minOccurs="0"/>
+ <xs:element name="distribution" type="phy:Distribution" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="date" type="phy:Date" minOccurs="0"/>
+ <xs:element name="reference" type="phy:Reference" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="clade" type="phy:Clade" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="branch_length" type="xs:double"/>
+ <xs:attribute name="id_source" type="phy:id_source"/>
+ <xs:attribute name="collapse" type="xs:boolean"/>
+ </xs:complexType>
+ <!-- Taxonomy:-->
+ <xs:complexType name="Taxonomy">
+ <xs:annotation>
+ <xs:documentation> Element Taxonomy is used to describe taxonomic information for a clade. Element 'code' is
+ intended to store UniProt/Swiss-Prot style organism codes (e.g. 'APLCA' for the California sea hare 'Aplysia
+ californica') or other styles of mnemonics (e.g. 'Aca'). Element 'authority' is used to keep the authority,
+ such as 'J. G. Cooper, 1863', associated with the 'scientific_name'. Element 'id' is used for a unique
+ identifier of a taxon (for example '6500' with 'ncbi_taxonomy' as 'provider' for the California sea hare).
+ Attribute 'id_source' is used to link other elements to a taxonomy (on the xml-level).</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="id" type="phy:Id" minOccurs="0"/>
+ <xs:element name="code" type="phy:TaxonomyCode" minOccurs="0"/>
+ <xs:element name="scientific_name" type="xs:token" minOccurs="0"/>
+ <xs:element name="authority" type="xs:token" minOccurs="0"/>
+ <xs:element name="common_name" type="xs:token" minOccurs="0"/>
+ <xs:element name="synonym" type="xs:token" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="rank" type="phy:Rank" minOccurs="0"/>
+ <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="id_source" type="phy:id_source"/>
+ </xs:complexType>
+ <xs:simpleType name="TaxonomyCode">
+ <xs:restriction base="xs:token">
+ <xs:pattern value="([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:simpleType name="Rank">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="domain"/>
+ <xs:enumeration value="superkingdom"/>
+ <xs:enumeration value="kingdom"/>
+ <xs:enumeration value="subkingdom"/>
+ <xs:enumeration value="branch"/>
+ <xs:enumeration value="infrakingdom"/>
+ <xs:enumeration value="superphylum"/>
+ <xs:enumeration value="phylum"/>
+ <xs:enumeration value="subphylum"/>
+ <xs:enumeration value="infraphylum"/>
+ <xs:enumeration value="microphylum"/>
+ <xs:enumeration value="superdivision"/>
+ <xs:enumeration value="division"/>
+ <xs:enumeration value="subdivision"/>
+ <xs:enumeration value="infradivision"/>
+ <xs:enumeration value="superclass"/>
+ <xs:enumeration value="class"/>
+ <xs:enumeration value="subclass"/>
+ <xs:enumeration value="infraclass"/>
+ <xs:enumeration value="superlegion"/>
+ <xs:enumeration value="legion"/>
+ <xs:enumeration value="sublegion"/>
+ <xs:enumeration value="infralegion"/>
+ <xs:enumeration value="supercohort"/>
+ <xs:enumeration value="cohort"/>
+ <xs:enumeration value="subcohort"/>
+ <xs:enumeration value="infracohort"/>
+ <xs:enumeration value="magnorder"/>
+ <xs:enumeration value="superorder"/>
+ <xs:enumeration value="order"/>
+ <xs:enumeration value="suborder"/>
+ <xs:enumeration value="infraorder"/>
+ <xs:enumeration value="superfamily"/>
+ <xs:enumeration value="family"/>
+ <xs:enumeration value="subfamily"/>
+ <xs:enumeration value="supertribe"/>
+ <xs:enumeration value="tribe"/>
+ <xs:enumeration value="subtribe"/>
+ <xs:enumeration value="infratribe"/>
+ <xs:enumeration value="genus"/>
+ <xs:enumeration value="subgenus"/>
+ <xs:enumeration value="superspecies"/>
+ <xs:enumeration value="species"/>
+ <xs:enumeration value="subspecies"/>
+ <xs:enumeration value="variety"/>
+ <xs:enumeration value="varietas"/>
+ <xs:enumeration value="subvariety"/>
+ <xs:enumeration value="form"/>
+ <xs:enumeration value="subform"/>
+ <xs:enumeration value="cultivar"/>
+ <xs:enumeration value="strain"/>
+ <xs:enumeration value="section"/>
+ <xs:enumeration value="subsection"/>
+ <xs:enumeration value="unknown"/>
+ <xs:enumeration value="other"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!-- Sequence:-->
+ <xs:complexType name="Sequence">
+ <xs:annotation>
+ <xs:documentation> Element Sequence is used to represent a molecular sequence (Protein, DNA, RNA) associated
+ with a node. 'symbol' is a short (maximal 20 characters) symbol of the sequence (e.g. 'ACTM') whereas
+ 'name' is used for the full name (e.g. 'muscle Actin'). 'gene_name' can be used when protein and gene names differ.
+ 'location' is used for the location of a sequence on a genome/chromosome. The actual sequence can be stored with the
+ 'mol_seq' element. Attribute 'type' is used to indicate the type of sequence ('dna', 'rna', or 'protein').
+ One intended use for 'id_ref' is to link a sequence to a taxonomy (via the taxonomy's 'id_source') in case
+ of multiple sequences and taxonomies per node. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="symbol" type="phy:SequenceSymbol" minOccurs="0"/>
+ <xs:element name="accession" type="phy:Accession" minOccurs="0"/>
+ <xs:element name="name" type="xs:token" minOccurs="0"/>
+ <xs:element name="gene_name" type="xs:token" minOccurs="0"/>
+ <xs:element name="location" type="xs:token" minOccurs="0"/>
+ <xs:element name="mol_seq" type="phy:MolSeq" minOccurs="0"/>
+ <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="annotation" type="phy:Annotation" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="cross_references" type="phy:CrossReferences" minOccurs="0"/>
+ <xs:element name="domain_architecture" type="phy:DomainArchitecture" minOccurs="0"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="type" type="phy:SequenceType"/>
+ <xs:attribute name="id_source" type="phy:id_source"/>
+ <xs:attribute name="id_ref" type="phy:id_ref"/>
+ </xs:complexType>
+ <xs:simpleType name="SequenceSymbol">
+ <xs:restriction base="xs:token">
+ <xs:pattern value="\S{1,20}"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:complexType name="MolSeq">
+ <xs:annotation>
+ <xs:documentation> Element 'mol_seq' is used to store molecular sequences. The 'is_aligned' attribute is used
+ to indicated that this molecular sequence is aligned with all other sequences in the same phylogeny for
+ which 'is aligned' is true as well (which, in most cases, means that gaps were introduced, and that all
+ sequences for which 'is aligned' is true must have the same length).</xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="is_aligned" type="xs:boolean"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <xs:simpleType name="SequenceType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="rna"/>
+ <xs:enumeration value="dna"/>
+ <xs:enumeration value="protein"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!-- Accession:-->
+ <xs:complexType name="Accession">
+ <xs:annotation>
+ <xs:documentation> Element Accession is used to capture the local part in a sequence identifier (e.g. 'P17304'
+ in 'UniProtKB:P17304', in which case the 'source' attribute would be 'UniProtKB'). </xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="source" type="xs:token" use="required"/>
+ <xs:attribute name="comment" type="xs:token"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- CrossReferences: -->
+ <xs:complexType name="CrossReferences">
+ <xs:annotation>
+ <xs:documentation> Used to store accessions to additional resources. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="accession" type="phy:Accession" minOccurs="1" maxOccurs="unbounded"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- DomainArchitecture: -->
+ <xs:complexType name="DomainArchitecture">
+ <xs:annotation>
+ <xs:documentation> This is used describe the domain architecture of a protein. Attribute 'length' is the total
+ length of the protein</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="domain" type="phy:ProteinDomain" minOccurs="1" maxOccurs="unbounded"/>
+ </xs:sequence>
+ <xs:attribute name="length" type="xs:nonNegativeInteger"/>
+ </xs:complexType>
+ <xs:complexType name="ProteinDomain">
+ <xs:annotation>
+ <xs:documentation> To represent an individual domain in a domain architecture. The name/unique identifier is
+ described via the 'id' attribute. 'confidence' can be used to store (i.e.) E-values.</xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="from" type="xs:nonNegativeInteger" use="required"/>
+ <xs:attribute name="to" type="xs:nonNegativeInteger" use="required"/>
+ <xs:attribute name="confidence" type="xs:double"/>
+ <xs:attribute name="id" type="xs:token"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Events:-->
+ <xs:complexType name="Events">
+ <xs:annotation>
+ <xs:documentation> Events at the root node of a clade (e.g. one gene duplication). </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="type" type="phy:EventType" minOccurs="0"/>
+ <xs:element name="duplications" type="xs:nonNegativeInteger" minOccurs="0"/>
+ <xs:element name="speciations" type="xs:nonNegativeInteger" minOccurs="0"/>
+ <xs:element name="losses" type="xs:nonNegativeInteger" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ </xs:sequence>
+ </xs:complexType>
+ <xs:simpleType name="EventType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="transfer"/>
+ <xs:enumeration value="fusion"/>
+ <xs:enumeration value="speciation_or_duplication"/>
+ <xs:enumeration value="other"/>
+ <xs:enumeration value="mixed"/>
+ <xs:enumeration value="unassigned"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!--BinaryCharacters:-->
+ <xs:complexType name="BinaryCharacters">
+ <xs:annotation>
+ <xs:documentation> The names and/or counts of binary characters present, gained, and lost at the root of a
+ clade. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="gained" type="phy:BinaryCharacterList" minOccurs="0"/>
+ <xs:element name="lost" type="phy:BinaryCharacterList" minOccurs="0"/>
+ <xs:element name="present" type="phy:BinaryCharacterList" minOccurs="0"/>
+ <xs:element name="absent" type="phy:BinaryCharacterList" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="type" type="xs:token"/>
+ <xs:attribute name="gained_count" type="xs:nonNegativeInteger"/>
+ <xs:attribute name="lost_count" type="xs:nonNegativeInteger"/>
+ <xs:attribute name="present_count" type="xs:nonNegativeInteger"/>
+ <xs:attribute name="absent_count" type="xs:nonNegativeInteger"/>
+ </xs:complexType>
+ <xs:complexType name="BinaryCharacterList">
+ <xs:sequence>
+ <xs:element name="bc" type="xs:token" maxOccurs="unbounded"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- Reference:-->
+ <xs:complexType name="Reference">
+ <xs:annotation>
+ <xs:documentation> A literature reference for a clade. It is recommended to use the 'doi' attribute instead of
+ the free text 'desc' element whenever possible. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="doi" type="xs:token"/>
+ </xs:complexType>
+ <!-- Annotation:-->
+ <xs:complexType name="Annotation">
+ <xs:annotation>
+ <xs:documentation> The annotation of a molecular sequence. It is recommended to annotate by using the optional
+ 'ref' attribute (some examples of acceptable values for the ref attribute: 'GO:0008270',
+ 'KEGG:Tetrachloroethene degradation', 'EC:1.1.1.1'). Optional element 'desc' allows for a free text
+ description. Optional element 'confidence' is used to state the type and value of support for a annotation.
+ Similarly, optional attribute 'evidence' is used to describe the evidence for a annotation as free text
+ (e.g. 'experimental'). Optional element 'property' allows for further, typed and referenced annotations from
+ external resources.</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>
+ </xs:sequence>
+ <xs:attribute name="ref" type="phy:ref"/>
+ <xs:attribute name="source" type="xs:token"/>
+ <xs:attribute name="evidence" type="xs:token"/>
+ <xs:attribute name="type" type="xs:token"/>
+ </xs:complexType>
+ <!-- Property:-->
+ <xs:complexType name="Property" mixed="true">
+ <xs:annotation>
+ <xs:documentation> Property allows for typed and referenced properties from external resources to be attached
+ to 'Phylogeny', 'Clade', and 'Annotation'. The value of a property is its mixed (free text) content.
+ Attribute 'datatype' indicates the type of a property and is limited to xsd-datatypes (e.g. 'xsd:string',
+ 'xsd:boolean', 'xsd:integer', 'xsd:decimal', 'xsd:float', 'xsd:double', 'xsd:date', 'xsd:anyURI'). Attribute
+ 'applies_to' indicates the item to which a property applies to (e.g. 'node' for the parent node of a clade,
+ 'parent_branch' for the parent branch of a clade). Attribute 'id_ref' allows to attached a property
+ specifically to one element (on the xml-level). Optional attribute 'unit' is used to indicate the unit of
+ the property. An example: <property datatype="xsd:integer" ref="NOAA:depth" applies_to="clade"
+ unit="METRIC:m"> 200 </property> </xs:documentation>
+ </xs:annotation>
+ <xs:attribute name="ref" type="phy:ref" use="required"/>
+ <xs:attribute name="unit" type="phy:ref"/>
+ <xs:attribute name="datatype" type="phy:PropertyDataType" use="required"/>
+ <xs:attribute name="applies_to" type="phy:AppliesTo" use="required"/>
+ <xs:attribute name="id_ref" type="phy:id_ref"/>
+ </xs:complexType>
+ <xs:simpleType name="ref">
+ <xs:restriction base="xs:token">
+ <xs:pattern value="[a-zA-Z0-9_]+:\S+"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:simpleType name="AppliesTo">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="phylogeny"/>
+ <xs:enumeration value="clade"/>
+ <xs:enumeration value="node"/>
+ <xs:enumeration value="annotation"/>
+ <xs:enumeration value="parent_branch"/>
+ <xs:enumeration value="other"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:simpleType name="PropertyDataType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="xsd:string"/>
+ <xs:enumeration value="xsd:boolean"/>
+ <xs:enumeration value="xsd:decimal"/>
+ <xs:enumeration value="xsd:float"/>
+ <xs:enumeration value="xsd:double"/>
+ <xs:enumeration value="xsd:duration"/>
+ <xs:enumeration value="xsd:dateTime"/>
+ <xs:enumeration value="xsd:time"/>
+ <xs:enumeration value="xsd:date"/>
+ <xs:enumeration value="xsd:gYearMonth"/>
+ <xs:enumeration value="xsd:gYear"/>
+ <xs:enumeration value="xsd:gMonthDay"/>
+ <xs:enumeration value="xsd:gDay"/>
+ <xs:enumeration value="xsd:gMonth"/>
+ <xs:enumeration value="xsd:hexBinary"/>
+ <xs:enumeration value="xsd:base64Binary"/>
+ <xs:enumeration value="xsd:anyURI"/>
+ <xs:enumeration value="xsd:normalizedString"/>
+ <xs:enumeration value="xsd:token"/>
+ <xs:enumeration value="xsd:integer"/>
+ <xs:enumeration value="xsd:nonPositiveInteger"/>
+ <xs:enumeration value="xsd:negativeInteger"/>
+ <xs:enumeration value="xsd:long"/>
+ <xs:enumeration value="xsd:int"/>
+ <xs:enumeration value="xsd:short"/>
+ <xs:enumeration value="xsd:byte"/>
+ <xs:enumeration value="xsd:nonNegativeInteger"/>
+ <xs:enumeration value="xsd:unsignedLong"/>
+ <xs:enumeration value="xsd:unsignedInt"/>
+ <xs:enumeration value="xsd:unsignedShort"/>
+ <xs:enumeration value="xsd:unsignedByte"/>
+ <xs:enumeration value="xsd:positiveInteger"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!--Uri-->
+ <xs:complexType name="Uri">
+ <xs:annotation>
+ <xs:documentation> A uniform resource identifier. In general, this is expected to be an URL (for example, to
+ link to an image on a website, in which case the 'type' attribute might be 'image' and 'desc' might be
+ 'image of a California sea hare'). </xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:anyURI">
+ <xs:attribute name="desc" type="xs:token"/>
+ <xs:attribute name="type" type="xs:token"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Confidence:-->
+ <xs:complexType name="Confidence">
+ <xs:annotation>
+ <xs:documentation> A general purpose confidence element. For example this can be used to express the bootstrap
+ support value of a clade (in which case the 'type' attribute is 'bootstrap').</xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:double">
+ <xs:attribute name="type" type="xs:token" use="required"/>
+ <xs:attribute name="stddev" type="xs:double"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Identifier:-->
+ <xs:complexType name="Id">
+ <xs:annotation>
+ <xs:documentation> A general purpose identifier element. Allows to indicate the provider (or authority) of an
+ identifier. </xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="provider" type="xs:token"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Distribution:-->
+ <xs:complexType name="Distribution">
+ <xs:annotation>
+ <xs:documentation> The geographic distribution of the items of a clade (species, sequences), intended for
+ phylogeographic applications. The location can be described either by free text in the 'desc' element and/or
+ by the coordinates of one or more 'Points' (similar to the 'Point' element in Google's KML format) or by
+ 'Polygons'. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ <xs:element name="point" type="phy:Point" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="polygon" type="phy:Polygon" minOccurs="0" maxOccurs="unbounded"/>
+ </xs:sequence>
+ </xs:complexType>
+ <xs:complexType name="Point">
+ <xs:annotation>
+ <xs:documentation> The coordinates of a point with an optional altitude (used by element 'Distribution').
+ Required attributes are the 'geodetic_datum' used to indicate the geodetic datum (also called 'map datum',
+ for example Google's KML uses 'WGS84'). Attribute 'alt_unit' is the unit for the altitude (e.g. 'meter').
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="lat" type="xs:decimal"/>
+ <xs:element name="long" type="xs:decimal"/>
+ <xs:element name="alt" type="xs:decimal" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="geodetic_datum" type="xs:token" use="required"/>
+ <xs:attribute name="alt_unit" type="xs:token"/>
+ </xs:complexType>
+ <xs:complexType name="Polygon">
+ <xs:annotation>
+ <xs:documentation> A polygon defined by a list of 'Points' (used by element 'Distribution').
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="point" type="phy:Point" minOccurs="3" maxOccurs="unbounded"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- Date:-->
+ <xs:complexType name="Date">
+ <xs:annotation>
+ <xs:documentation> A date associated with a clade/node. Its value can be numerical by using the 'value' element
+ and/or free text with the 'desc' element' (e.g. 'Silurian'). If a numerical value is used, it is recommended
+ to employ the 'unit' attribute to indicate the type of the numerical value (e.g. 'mya' for 'million years
+ ago'). The elements 'minimum' and 'maximum' are used the indicate a range/confidence
+ interval</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ <xs:element name="value" type="xs:decimal" minOccurs="0"/>
+ <xs:element name="minimum" type="xs:decimal" minOccurs="0"/>
+ <xs:element name="maximum" type="xs:decimal" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="unit" type="xs:token"/>
+ </xs:complexType>
+ <!-- BranchColor:-->
+ <xs:complexType name="BranchColor">
+ <xs:annotation>
+ <xs:documentation> This indicates the color of a clade when rendered (the color applies to the whole clade
+ unless overwritten by the color(s) of sub clades).</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="red" type="xs:unsignedByte"/>
+ <xs:element name="green" type="xs:unsignedByte"/>
+ <xs:element name="blue" type="xs:unsignedByte"/>
+ <xs:element name="alpha" type="xs:unsignedByte" minOccurs="0"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- SequenceRelation:-->
+ <xs:complexType name="SequenceRelation">
+ <xs:annotation>
+ <xs:documentation> This is used to express a typed relationship between two sequences. For example it could be
+ used to describe an orthology (in which case attribute 'type' is 'orthology'). </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>
+ <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>
+ <xs:attribute name="distance" type="xs:double"/>
+ <xs:attribute name="type" type="phy:SequenceRelationType" use="required"/>
+ </xs:complexType>
+ <xs:simpleType name="SequenceRelationType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="orthology"/>
+ <xs:enumeration value="one_to_one_orthology"/>
+ <xs:enumeration value="super_orthology"/>
+ <xs:enumeration value="paralogy"/>
+ <xs:enumeration value="ultra_paralogy"/>
+ <xs:enumeration value="xenology"/>
+ <xs:enumeration value="unknown"/>
+ <xs:enumeration value="other"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!-- CladeRelation:-->
+ <xs:complexType name="CladeRelation">
+ <xs:annotation>
+ <xs:documentation> This is used to express a typed relationship between two clades. For example it could be
+ used to describe multiple parents of a clade.</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>
+ <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>
+ <xs:attribute name="distance" type="xs:double"/>
+ <xs:attribute name="type" type="xs:token" use="required"/>
+ </xs:complexType>
+ <!-- Used to link elements together on the xml level:-->
+ <xs:simpleType name="id_source">
+ <xs:restriction base="xs:ID"/>
+ </xs:simpleType>
+ <xs:simpleType name="id_ref">
+ <xs:restriction base="xs:IDREF"/>
+ </xs:simpleType>
+</xs:schema>
final static boolean __ALLOW_PHYLOGENETIC_INFERENCE = true;
public final static String PRG_NAME = "Archaeopteryx";
final static String VERSION = "0.9914 test";
- final static String PRG_DATE = "160713";
+ final static String PRG_DATE = "160714";
final static String DEFAULT_CONFIGURATION_FILE_NAME = "_aptx_configuration_file";
final static String[] DEFAULT_FONT_CHOICES = {
"Arial Unicode MS", "Dialog", "SansSerif", "Sans", "Arial", "Helvetica" };
private final static String PATH_TO_TEST_DATA = System.getProperty( "user.dir" )
+ ForesterUtil.getFileSeparator() + "test_data"
+ ForesterUtil.getFileSeparator();
- private final static boolean PERFORM_DB_TESTS = false;
- private static final boolean PERFORM_WEB_TREE_ACCESS = false;
+ private final static boolean PERFORM_DB_TESTS = true;
+ private static final boolean PERFORM_WEB_TREE_ACCESS = true;
private static final String PHYLOXML_LOCAL_XSD = PATH_TO_RESOURCES + "phyloxml_schema/"
+ ForesterConstants.PHYLO_XML_VERSION + "/"
+ ForesterConstants.PHYLO_XML_XSD;
public final static String FORESTER_VERSION = "1.042";
public final static String FORESTER_DATE = "160712";
- public final static String PHYLO_XML_VERSION = "1.10";
+ public final static String PHYLO_XML_VERSION = "1.20";
public final static String PHYLO_XML_LOCATION = "http://www.phyloxml.org";
public final static String PHYLO_XML_XSD = "phyloxml.xsd";
public final static String XML_SCHEMA_INSTANCE = "http://www.w3.org/2001/XMLSchema-instance";
-<?xml version="1.0" encoding="UTF-8"?>\r
-<!-- -->\r
-<!-- phyloXML -->\r
-<!-- -->\r
-<!-- schema in XMLSchema -->\r
-<!-- -->\r
-<!-- License: The phyloXML XML Schema Definition is -->\r
-<!-- dual-licensed under the LGPL or Ruby's -->\r
-<!-- License. -->\r
-<!-- You can redistribute and/or modify -->\r
-<!-- it under either the terms of the LGPL, -->\r
-<!-- or Ruby's License. -->\r
-<!-- see: http://www.ruby-lang.org/en/about/license.txt -->\r
-<!-- -->\r
-<!-- Copyright (c) 2008-2013 Christian Zmasek -->\r
-<!-- -->\r
-<!-- www.phyloxml.org -->\r
-<!-- Version: 1.16 -->\r
-<!-- Last modified: 2013.09.30 by Christian Zmasek -->\r
-<!-- -->\r
-<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:phy="http://www.phyloxml.org"\r
- targetNamespace="http://www.phyloxml.org" elementFormDefault="qualified" attributeFormDefault="unqualified">\r
- <xs:annotation>\r
- <xs:documentation> phyloXML is an XML language to describe evolutionary trees and associated data. Version: 1.10.\r
- License: dual-licensed under the LGPL or Ruby's License. Copyright (c) 2008-2011 Christian M Zmasek.</xs:documentation>\r
- </xs:annotation>\r
- <!-- phyloxml is the root element:-->\r
- <xs:element name="phyloxml" type="phy:Phyloxml"/>\r
- <!-- phyloXML definition:-->\r
- <xs:complexType name="Phyloxml">\r
- <xs:annotation>\r
- <xs:documentation> 'phyloxml' is the name of the root element. Phyloxml contains an arbitrary number of\r
- 'phylogeny' elements (each representing one phylogeny) possibly followed by elements from other namespaces.\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence maxOccurs="unbounded">\r
- <xs:element name="phylogeny" type="phy:Phylogeny" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- Phylogeny:-->\r
- <xs:complexType name="Phylogeny">\r
- <xs:annotation>\r
- <xs:documentation> Element Phylogeny is used to represent a phylogeny. The required attribute 'rooted' is used\r
- to indicate whether the phylogeny is rooted or not. The attribute 'rerootable' can be used to indicate that\r
- the phylogeny is not allowed to be rooted differently (i.e. because it is associated with root dependent\r
- data, such as gene duplications). The attribute 'type' can be used to indicate the type of phylogeny (i.e.\r
- 'gene tree'). It is recommended to use the attribute 'branch_length_unit' if the phylogeny has branch\r
- lengths. Element clade is used in a recursive manner to describe the topology of a phylogenetic\r
- tree.</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="id" type="phy:Id" minOccurs="0"/>\r
- <xs:element name="description" type="xs:token" minOccurs="0"/>\r
- <xs:element name="date" type="xs:dateTime" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="clade" type="phy:Clade" minOccurs="0"/>\r
- <xs:element name="clade_relation" type="phy:CladeRelation" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="sequence_relation" type="phy:SequenceRelation" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="rooted" type="xs:boolean" use="required"/>\r
- <xs:attribute name="rerootable" type="xs:boolean"/>\r
- <xs:attribute name="branch_length_unit" type="xs:token"/>\r
- <xs:attribute name="type" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- Clade:-->\r
- <xs:complexType name="Clade">\r
- <xs:annotation>\r
- <xs:documentation> Element Clade is used in a recursive manner to describe the topology of a phylogenetic tree.\r
- The parent branch length of a clade can be described either with the 'branch_length' element or the\r
- 'branch_length' attribute (it is not recommended to use both at the same time, though). Usage of the\r
- 'branch_length' attribute allows for a less verbose description. Element 'confidence' is used to indicate\r
- the support for a clade/parent branch. Element 'events' is used to describe such events as gene-duplications\r
- at the root node/parent branch of a clade. Element 'width' is the branch width for this clade (including\r
- parent branch). Both 'color' and 'width' elements apply for the whole clade unless overwritten in-sub\r
- clades. Attribute 'id_source' is used to link other elements to a clade (on the xml-level).\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="branch_length" type="xs:double" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="width" type="xs:double" minOccurs="0"/>\r
- <xs:element name="color" type="phy:BranchColor" minOccurs="0"/>\r
- <xs:element name="taxonomy" type="phy:Taxonomy" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="sequence" type="phy:Sequence" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="events" type="phy:Events" minOccurs="0"/>\r
- <xs:element name="binary_characters" type="phy:BinaryCharacters" minOccurs="0"/>\r
- <xs:element name="distribution" type="phy:Distribution" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="date" type="phy:Date" minOccurs="0"/>\r
- <xs:element name="reference" type="phy:Reference" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="clade" type="phy:Clade" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="branch_length" type="xs:double"/>\r
- <xs:attribute name="id_source" type="phy:id_source"/>\r
- <xs:attribute name="collapse" type="xs:boolean"/>\r
- </xs:complexType>\r
- <!-- Taxonomy:-->\r
- <xs:complexType name="Taxonomy">\r
- <xs:annotation>\r
- <xs:documentation> Element Taxonomy is used to describe taxonomic information for a clade. Element 'code' is\r
- intended to store UniProt/Swiss-Prot style organism codes (e.g. 'APLCA' for the California sea hare 'Aplysia\r
- californica') or other styles of mnemonics (e.g. 'Aca'). Element 'authority' is used to keep the authority,\r
- such as 'J. G. Cooper, 1863', associated with the 'scientific_name'. Element 'id' is used for a unique\r
- identifier of a taxon (for example '6500' with 'ncbi_taxonomy' as 'provider' for the California sea hare).\r
- Attribute 'id_source' is used to link other elements to a taxonomy (on the xml-level).</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="id" type="phy:Id" minOccurs="0"/>\r
- <xs:element name="code" type="phy:TaxonomyCode" minOccurs="0"/>\r
- <xs:element name="scientific_name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="authority" type="xs:token" minOccurs="0"/>\r
- <xs:element name="common_name" type="xs:token" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="synonym" type="xs:token" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="rank" type="phy:Rank" minOccurs="0"/>\r
- <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="id_source" type="phy:id_source"/>\r
- </xs:complexType>\r
- <xs:simpleType name="TaxonomyCode">\r
- <xs:restriction base="xs:token">\r
- <xs:pattern value="([A-Z9][A-Z]{2}[A-Z0-9]{2})|RAT|PIG|PEA"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:simpleType name="Rank">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="domain"/>\r
- <xs:enumeration value="superkingdom"/>\r
- <xs:enumeration value="kingdom"/>\r
- <xs:enumeration value="subkingdom"/>\r
- <xs:enumeration value="branch"/>\r
- <xs:enumeration value="infrakingdom"/>\r
- <xs:enumeration value="superphylum"/>\r
- <xs:enumeration value="phylum"/>\r
- <xs:enumeration value="subphylum"/>\r
- <xs:enumeration value="infraphylum"/>\r
- <xs:enumeration value="microphylum"/>\r
- <xs:enumeration value="superdivision"/>\r
- <xs:enumeration value="division"/>\r
- <xs:enumeration value="subdivision"/>\r
- <xs:enumeration value="infradivision"/>\r
- <xs:enumeration value="superclass"/>\r
- <xs:enumeration value="class"/>\r
- <xs:enumeration value="subclass"/>\r
- <xs:enumeration value="infraclass"/>\r
- <xs:enumeration value="superlegion"/>\r
- <xs:enumeration value="legion"/>\r
- <xs:enumeration value="sublegion"/>\r
- <xs:enumeration value="infralegion"/>\r
- <xs:enumeration value="supercohort"/>\r
- <xs:enumeration value="cohort"/>\r
- <xs:enumeration value="subcohort"/>\r
- <xs:enumeration value="infracohort"/>\r
- <xs:enumeration value="magnorder"/>\r
- <xs:enumeration value="superorder"/>\r
- <xs:enumeration value="order"/>\r
- <xs:enumeration value="suborder"/>\r
- <xs:enumeration value="infraorder"/>\r
- <xs:enumeration value="superfamily"/>\r
- <xs:enumeration value="family"/>\r
- <xs:enumeration value="subfamily"/>\r
- <xs:enumeration value="supertribe"/>\r
- <xs:enumeration value="tribe"/>\r
- <xs:enumeration value="subtribe"/>\r
- <xs:enumeration value="infratribe"/>\r
- <xs:enumeration value="genus"/>\r
- <xs:enumeration value="subgenus"/>\r
- <xs:enumeration value="superspecies"/>\r
- <xs:enumeration value="species"/>\r
- <xs:enumeration value="subspecies"/>\r
- <xs:enumeration value="variety"/>\r
- <xs:enumeration value="varietas"/>\r
- <xs:enumeration value="subvariety"/>\r
- <xs:enumeration value="form"/>\r
- <xs:enumeration value="subform"/>\r
- <xs:enumeration value="cultivar"/>\r
- <xs:enumeration value="strain"/>\r
- <xs:enumeration value="section"/>\r
- <xs:enumeration value="subsection"/>\r
- <xs:enumeration value="unknown"/>\r
- <xs:enumeration value="other"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!-- Sequence:-->\r
- <xs:complexType name="Sequence">\r
- <xs:annotation>\r
- <xs:documentation> Element Sequence is used to represent a molecular sequence (Protein, DNA, RNA) associated\r
- with a node. 'symbol' is a short (maximal 20 characters) symbol of the sequence (e.g. 'ACTM') whereas\r
- 'name' is used for the full name (e.g. 'muscle Actin'). 'gene_name' can be used when protein and gene names differ.\r
- 'location' is used for the location of a sequence on a genome/chromosome. The actual sequence can be stored with the \r
- 'mol_seq' element. Attribute 'type' is used to indicate the type of sequence ('dna', 'rna', or 'protein'). \r
- One intended use for 'id_ref' is to link a sequence to a taxonomy (via the taxonomy's 'id_source') in case \r
- of multiple sequences and taxonomies per node. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="symbol" type="phy:SequenceSymbol" minOccurs="0"/>\r
- <xs:element name="accession" type="phy:Accession" minOccurs="0"/>\r
- <xs:element name="name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="gene_name" type="xs:token" minOccurs="0"/>\r
- <xs:element name="location" type="xs:token" minOccurs="0"/>\r
- <xs:element name="mol_seq" type="phy:MolSeq" minOccurs="0"/>\r
- <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="annotation" type="phy:Annotation" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="cross_references" type="phy:CrossReferences" minOccurs="0"/>\r
- <xs:element name="domain_architecture" type="phy:DomainArchitecture" minOccurs="0"/>\r
- <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>\r
- </xs:sequence>\r
- <xs:attribute name="type" type="phy:SequenceType"/>\r
- <xs:attribute name="id_source" type="phy:id_source"/>\r
- <xs:attribute name="id_ref" type="phy:id_ref"/>\r
- </xs:complexType>\r
- <xs:simpleType name="SequenceSymbol">\r
- <xs:restriction base="xs:token">\r
- <xs:pattern value="\S{1,20}"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:complexType name="MolSeq">\r
- <xs:annotation>\r
- <xs:documentation> Element 'mol_seq' is used to store molecular sequences. The 'is_aligned' attribute is used\r
- to indicated that this molecular sequence is aligned with all other sequences in the same phylogeny for\r
- which 'is aligned' is true as well (which, in most cases, means that gaps were introduced, and that all\r
- sequences for which 'is aligned' is true must have the same length).</xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="is_aligned" type="xs:boolean"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <xs:simpleType name="SequenceType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="rna"/>\r
- <xs:enumeration value="dna"/>\r
- <xs:enumeration value="protein"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!-- Accession:-->\r
- <xs:complexType name="Accession">\r
- <xs:annotation>\r
- <xs:documentation> Element Accession is used to capture the local part in a sequence identifier (e.g. 'P17304'\r
- in 'UniProtKB:P17304', in which case the 'source' attribute would be 'UniProtKB'). </xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="source" type="xs:token" use="required"/>\r
- <xs:attribute name="comment" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- CrossReferences: -->\r
- <xs:complexType name="CrossReferences">\r
- <xs:annotation>\r
- <xs:documentation> Used to store accessions to additional resources. </xs:documentation>\r
- </xs:annotation> \r
- <xs:sequence>\r
- <xs:element name="accession" type="phy:Accession" minOccurs="1" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType> \r
- <!-- DomainArchitecture: -->\r
- <xs:complexType name="DomainArchitecture">\r
- <xs:annotation>\r
- <xs:documentation> This is used describe the domain architecture of a protein. Attribute 'length' is the total\r
- length of the protein</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="domain" type="phy:ProteinDomain" minOccurs="1" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- <xs:attribute name="length" type="xs:nonNegativeInteger"/>\r
- </xs:complexType>\r
- <xs:complexType name="ProteinDomain">\r
- <xs:annotation>\r
- <xs:documentation> To represent an individual domain in a domain architecture. The name/unique identifier is\r
- described via the 'id' attribute. 'confidence' can be used to store (i.e.) E-values.</xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="from" type="xs:nonNegativeInteger" use="required"/>\r
- <xs:attribute name="to" type="xs:nonNegativeInteger" use="required"/>\r
- <xs:attribute name="confidence" type="xs:double"/>\r
- <xs:attribute name="id" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Events:-->\r
- <xs:complexType name="Events">\r
- <xs:annotation>\r
- <xs:documentation> Events at the root node of a clade (e.g. one gene duplication). </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="type" type="phy:EventType" minOccurs="0"/>\r
- <xs:element name="duplications" type="xs:nonNegativeInteger" minOccurs="0"/>\r
- <xs:element name="speciations" type="xs:nonNegativeInteger" minOccurs="0"/>\r
- <xs:element name="losses" type="xs:nonNegativeInteger" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <xs:simpleType name="EventType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="transfer"/>\r
- <xs:enumeration value="fusion"/>\r
- <xs:enumeration value="speciation_or_duplication"/>\r
- <xs:enumeration value="other"/>\r
- <xs:enumeration value="mixed"/>\r
- <xs:enumeration value="unassigned"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!--BinaryCharacters:-->\r
- <xs:complexType name="BinaryCharacters">\r
- <xs:annotation>\r
- <xs:documentation> The names and/or counts of binary characters present, gained, and lost at the root of a\r
- clade. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="gained" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- <xs:element name="lost" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- <xs:element name="present" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- <xs:element name="absent" type="phy:BinaryCharacterList" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="type" type="xs:token"/>\r
- <xs:attribute name="gained_count" type="xs:nonNegativeInteger"/>\r
- <xs:attribute name="lost_count" type="xs:nonNegativeInteger"/>\r
- <xs:attribute name="present_count" type="xs:nonNegativeInteger"/>\r
- <xs:attribute name="absent_count" type="xs:nonNegativeInteger"/>\r
- </xs:complexType>\r
- <xs:complexType name="BinaryCharacterList">\r
- <xs:sequence>\r
- <xs:element name="bc" type="xs:token" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- Reference:-->\r
- <xs:complexType name="Reference">\r
- <xs:annotation>\r
- <xs:documentation> A literature reference for a clade. It is recommended to use the 'doi' attribute instead of\r
- the free text 'desc' element whenever possible. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="doi" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- Annotation:-->\r
- <xs:complexType name="Annotation">\r
- <xs:annotation>\r
- <xs:documentation> The annotation of a molecular sequence. It is recommended to annotate by using the optional\r
- 'ref' attribute (some examples of acceptable values for the ref attribute: 'GO:0008270',\r
- 'KEGG:Tetrachloroethene degradation', 'EC:1.1.1.1'). Optional element 'desc' allows for a free text\r
- description. Optional element 'confidence' is used to state the type and value of support for a annotation.\r
- Similarly, optional attribute 'evidence' is used to describe the evidence for a annotation as free text\r
- (e.g. 'experimental'). Optional element 'property' allows for further, typed and referenced annotations from\r
- external resources.</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="uri" type="phy:Uri" minOccurs="0" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- <xs:attribute name="ref" type="phy:ref"/>\r
- <xs:attribute name="source" type="xs:token"/>\r
- <xs:attribute name="evidence" type="xs:token"/>\r
- <xs:attribute name="type" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- Property:-->\r
- <xs:complexType name="Property" mixed="true">\r
- <xs:annotation>\r
- <xs:documentation> Property allows for typed and referenced properties from external resources to be attached\r
- to 'Phylogeny', 'Clade', and 'Annotation'. The value of a property is its mixed (free text) content.\r
- Attribute 'datatype' indicates the type of a property and is limited to xsd-datatypes (e.g. 'xsd:string',\r
- 'xsd:boolean', 'xsd:integer', 'xsd:decimal', 'xsd:float', 'xsd:double', 'xsd:date', 'xsd:anyURI'). Attribute\r
- 'applies_to' indicates the item to which a property applies to (e.g. 'node' for the parent node of a clade,\r
- 'parent_branch' for the parent branch of a clade). Attribute 'id_ref' allows to attached a property\r
- specifically to one element (on the xml-level). Optional attribute 'unit' is used to indicate the unit of\r
- the property. An example: <property datatype="xsd:integer" ref="NOAA:depth" applies_to="clade"\r
- unit="METRIC:m"> 200 </property> </xs:documentation>\r
- </xs:annotation>\r
- <xs:attribute name="ref" type="phy:ref" use="required"/>\r
- <xs:attribute name="unit" type="phy:ref"/>\r
- <xs:attribute name="datatype" type="phy:PropertyDataType" use="required"/>\r
- <xs:attribute name="applies_to" type="phy:AppliesTo" use="required"/>\r
- <xs:attribute name="id_ref" type="phy:id_ref"/>\r
- </xs:complexType>\r
- <xs:simpleType name="ref">\r
- <xs:restriction base="xs:token">\r
- <xs:pattern value="[a-zA-Z0-9_]+:\S+"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:simpleType name="AppliesTo">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="phylogeny"/>\r
- <xs:enumeration value="clade"/>\r
- <xs:enumeration value="node"/>\r
- <xs:enumeration value="annotation"/>\r
- <xs:enumeration value="parent_branch"/>\r
- <xs:enumeration value="other"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <xs:simpleType name="PropertyDataType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="xsd:string"/>\r
- <xs:enumeration value="xsd:boolean"/>\r
- <xs:enumeration value="xsd:decimal"/>\r
- <xs:enumeration value="xsd:float"/>\r
- <xs:enumeration value="xsd:double"/>\r
- <xs:enumeration value="xsd:duration"/>\r
- <xs:enumeration value="xsd:dateTime"/>\r
- <xs:enumeration value="xsd:time"/>\r
- <xs:enumeration value="xsd:date"/>\r
- <xs:enumeration value="xsd:gYearMonth"/>\r
- <xs:enumeration value="xsd:gYear"/>\r
- <xs:enumeration value="xsd:gMonthDay"/>\r
- <xs:enumeration value="xsd:gDay"/>\r
- <xs:enumeration value="xsd:gMonth"/>\r
- <xs:enumeration value="xsd:hexBinary"/>\r
- <xs:enumeration value="xsd:base64Binary"/>\r
- <xs:enumeration value="xsd:anyURI"/>\r
- <xs:enumeration value="xsd:normalizedString"/>\r
- <xs:enumeration value="xsd:token"/>\r
- <xs:enumeration value="xsd:integer"/>\r
- <xs:enumeration value="xsd:nonPositiveInteger"/>\r
- <xs:enumeration value="xsd:negativeInteger"/>\r
- <xs:enumeration value="xsd:long"/>\r
- <xs:enumeration value="xsd:int"/>\r
- <xs:enumeration value="xsd:short"/>\r
- <xs:enumeration value="xsd:byte"/>\r
- <xs:enumeration value="xsd:nonNegativeInteger"/>\r
- <xs:enumeration value="xsd:unsignedLong"/>\r
- <xs:enumeration value="xsd:unsignedInt"/>\r
- <xs:enumeration value="xsd:unsignedShort"/>\r
- <xs:enumeration value="xsd:unsignedByte"/>\r
- <xs:enumeration value="xsd:positiveInteger"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!--Uri-->\r
- <xs:complexType name="Uri">\r
- <xs:annotation>\r
- <xs:documentation> A uniform resource identifier. In general, this is expected to be an URL (for example, to\r
- link to an image on a website, in which case the 'type' attribute might be 'image' and 'desc' might be\r
- 'image of a California sea hare'). </xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:anyURI">\r
- <xs:attribute name="desc" type="xs:token"/>\r
- <xs:attribute name="type" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Confidence:-->\r
- <xs:complexType name="Confidence">\r
- <xs:annotation>\r
- <xs:documentation> A general purpose confidence element. For example this can be used to express the bootstrap\r
- support value of a clade (in which case the 'type' attribute is 'bootstrap').</xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:double">\r
- <xs:attribute name="type" type="xs:token" use="required"/>\r
- <xs:attribute name="stddev" type="xs:double"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Identifier:-->\r
- <xs:complexType name="Id">\r
- <xs:annotation>\r
- <xs:documentation> A general purpose identifier element. Allows to indicate the provider (or authority) of an\r
- identifier. </xs:documentation>\r
- </xs:annotation>\r
- <xs:simpleContent>\r
- <xs:extension base="xs:token">\r
- <xs:attribute name="provider" type="xs:token"/>\r
- </xs:extension>\r
- </xs:simpleContent>\r
- </xs:complexType>\r
- <!-- Distribution:-->\r
- <xs:complexType name="Distribution">\r
- <xs:annotation>\r
- <xs:documentation> The geographic distribution of the items of a clade (species, sequences), intended for\r
- phylogeographic applications. The location can be described either by free text in the 'desc' element and/or\r
- by the coordinates of one or more 'Points' (similar to the 'Point' element in Google's KML format) or by\r
- 'Polygons'. </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- <xs:element name="point" type="phy:Point" minOccurs="0" maxOccurs="unbounded"/>\r
- <xs:element name="polygon" type="phy:Polygon" minOccurs="0" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <xs:complexType name="Point">\r
- <xs:annotation>\r
- <xs:documentation> The coordinates of a point with an optional altitude (used by element 'Distribution').\r
- Required attributes are the 'geodetic_datum' used to indicate the geodetic datum (also called 'map datum',\r
- for example Google's KML uses 'WGS84'). Attribute 'alt_unit' is the unit for the altitude (e.g. 'meter').\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="lat" type="xs:decimal"/>\r
- <xs:element name="long" type="xs:decimal"/>\r
- <xs:element name="alt" type="xs:decimal" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="geodetic_datum" type="xs:token" use="required"/>\r
- <xs:attribute name="alt_unit" type="xs:token"/>\r
- </xs:complexType>\r
- <xs:complexType name="Polygon">\r
- <xs:annotation>\r
- <xs:documentation> A polygon defined by a list of 'Points' (used by element 'Distribution').\r
- </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="point" type="phy:Point" minOccurs="3" maxOccurs="unbounded"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- Date:-->\r
- <xs:complexType name="Date">\r
- <xs:annotation>\r
- <xs:documentation> A date associated with a clade/node. Its value can be numerical by using the 'value' element\r
- and/or free text with the 'desc' element' (e.g. 'Silurian'). If a numerical value is used, it is recommended\r
- to employ the 'unit' attribute to indicate the type of the numerical value (e.g. 'mya' for 'million years\r
- ago'). The elements 'minimum' and 'maximum' are used the indicate a range/confidence\r
- interval</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="desc" type="xs:token" minOccurs="0"/>\r
- <xs:element name="value" type="xs:decimal" minOccurs="0"/>\r
- <xs:element name="minimum" type="xs:decimal" minOccurs="0"/>\r
- <xs:element name="maximum" type="xs:decimal" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="unit" type="xs:token"/>\r
- </xs:complexType>\r
- <!-- BranchColor:-->\r
- <xs:complexType name="BranchColor">\r
- <xs:annotation>\r
- <xs:documentation> This indicates the color of a clade when rendered (the color applies to the whole clade\r
- unless overwritten by the color(s) of sub clades).</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="red" type="xs:unsignedByte"/>\r
- <xs:element name="green" type="xs:unsignedByte"/>\r
- <xs:element name="blue" type="xs:unsignedByte"/>\r
- </xs:sequence>\r
- </xs:complexType>\r
- <!-- SequenceRelation:-->\r
- <xs:complexType name="SequenceRelation">\r
- <xs:annotation>\r
- <xs:documentation> This is used to express a typed relationship between two sequences. For example it could be\r
- used to describe an orthology (in which case attribute 'type' is 'orthology'). </xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="distance" type="xs:double"/>\r
- <xs:attribute name="type" type="phy:SequenceRelationType" use="required"/>\r
- </xs:complexType>\r
- <xs:simpleType name="SequenceRelationType">\r
- <xs:restriction base="xs:token">\r
- <xs:enumeration value="orthology"/>\r
- <xs:enumeration value="one_to_one_orthology"/>\r
- <xs:enumeration value="super_orthology"/>\r
- <xs:enumeration value="paralogy"/>\r
- <xs:enumeration value="ultra_paralogy"/>\r
- <xs:enumeration value="xenology"/>\r
- <xs:enumeration value="unknown"/>\r
- <xs:enumeration value="other"/>\r
- </xs:restriction>\r
- </xs:simpleType>\r
- <!-- CladeRelation:-->\r
- <xs:complexType name="CladeRelation">\r
- <xs:annotation>\r
- <xs:documentation> This is used to express a typed relationship between two clades. For example it could be\r
- used to describe multiple parents of a clade.</xs:documentation>\r
- </xs:annotation>\r
- <xs:sequence>\r
- <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>\r
- </xs:sequence>\r
- <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>\r
- <xs:attribute name="distance" type="xs:double"/>\r
- <xs:attribute name="type" type="xs:token" use="required"/>\r
- </xs:complexType>\r
- <!-- Used to link elements together on the xml level:-->\r
- <xs:simpleType name="id_source">\r
- <xs:restriction base="xs:ID"/>\r
- </xs:simpleType>\r
- <xs:simpleType name="id_ref">\r
- <xs:restriction base="xs:IDREF"/>\r
- </xs:simpleType>\r
-</xs:schema>\r
+<?xml version="1.0" encoding="UTF-8"?>
+<!-- -->
+<!-- phyloXML -->
+<!-- -->
+<!-- schema in XMLSchema -->
+<!-- -->
+<!-- License: The phyloXML XML Schema Definition is -->
+<!-- dual-licensed under the LGPL or Ruby's -->
+<!-- License. -->
+<!-- You can redistribute and/or modify -->
+<!-- it under either the terms of the LGPL, -->
+<!-- or Ruby's License. -->
+<!-- see: http://www.ruby-lang.org/en/about/license.txt -->
+<!-- -->
+<!-- Copyright (c) 2008-2011 Christian M Zmasek -->
+<!-- -->
+<!-- www.phyloxml.org -->
+<!-- Version: 1.10 -->
+<!-- Last modified: 2010.10.01 by Christian M Zmasek -->
+<!-- -->
+<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" xmlns:phy="http://www.phyloxml.org"
+ targetNamespace="http://www.phyloxml.org" elementFormDefault="qualified" attributeFormDefault="unqualified">
+ <xs:annotation>
+ <xs:documentation> phyloXML is an XML language to describe evolutionary trees and associated data. Version: 1.10.
+ License: dual-licensed under the LGPL or Ruby's License. Copyright (c) 2008-2011 Christian M Zmasek.</xs:documentation>
+ </xs:annotation>
+ <!-- phyloxml is the root element:-->
+ <xs:element name="phyloxml" type="phy:Phyloxml"/>
+ <!-- phyloXML definition:-->
+ <xs:complexType name="Phyloxml">
+ <xs:annotation>
+ <xs:documentation> 'phyloxml' is the name of the root element. Phyloxml contains an arbitrary number of
+ 'phylogeny' elements (each representing one phylogeny) possibly followed by elements from other namespaces.
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence maxOccurs="unbounded">
+ <xs:element name="phylogeny" type="phy:Phylogeny" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- Phylogeny:-->
+ <xs:complexType name="Phylogeny">
+ <xs:annotation>
+ <xs:documentation> Element Phylogeny is used to represent a phylogeny. The required attribute 'rooted' is used
+ to indicate whether the phylogeny is rooted or not. The attribute 'rerootable' can be used to indicate that
+ the phylogeny is not allowed to be rooted differently (i.e. because it is associated with root dependent
+ data, such as gene duplications). The attribute 'type' can be used to indicate the type of phylogeny (i.e.
+ 'gene tree'). It is recommended to use the attribute 'branch_length_unit' if the phylogeny has branch
+ lengths. Element clade is used in a recursive manner to describe the topology of a phylogenetic
+ tree.</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="name" type="xs:token" minOccurs="0"/>
+ <xs:element name="id" type="phy:Id" minOccurs="0"/>
+ <xs:element name="description" type="xs:token" minOccurs="0"/>
+ <xs:element name="date" type="xs:dateTime" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="clade" type="phy:Clade" minOccurs="0"/>
+ <xs:element name="clade_relation" type="phy:CladeRelation" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="sequence_relation" type="phy:SequenceRelation" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="rooted" type="xs:boolean" use="required"/>
+ <xs:attribute name="rerootable" type="xs:boolean"/>
+ <xs:attribute name="branch_length_unit" type="xs:token"/>
+ <xs:attribute name="type" type="xs:token"/>
+ </xs:complexType>
+ <!-- Clade:-->
+ <xs:complexType name="Clade">
+ <xs:annotation>
+ <xs:documentation> Element Clade is used in a recursive manner to describe the topology of a phylogenetic tree.
+ The parent branch length of a clade can be described either with the 'branch_length' element or the
+ 'branch_length' attribute (it is not recommended to use both at the same time, though). Usage of the
+ 'branch_length' attribute allows for a less verbose description. Element 'confidence' is used to indicate
+ the support for a clade/parent branch. Element 'events' is used to describe such events as gene-duplications
+ at the root node/parent branch of a clade. Element 'width' is the branch width for this clade (including
+ parent branch). Both 'color' and 'width' elements apply for the whole clade unless overwritten in-sub
+ clades. Attribute 'id_source' is used to link other elements to a clade (on the xml-level).
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="name" type="xs:token" minOccurs="0"/>
+ <xs:element name="branch_length" type="xs:double" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="width" type="xs:double" minOccurs="0"/>
+ <xs:element name="color" type="phy:BranchColor" minOccurs="0"/>
+ <xs:element name="node_id" type="phy:Id" minOccurs="0"/>
+ <xs:element name="taxonomy" type="phy:Taxonomy" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="sequence" type="phy:Sequence" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="events" type="phy:Events" minOccurs="0"/>
+ <xs:element name="binary_characters" type="phy:BinaryCharacters" minOccurs="0"/>
+ <xs:element name="distribution" type="phy:Distribution" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="date" type="phy:Date" minOccurs="0"/>
+ <xs:element name="reference" type="phy:Reference" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="clade" type="phy:Clade" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="branch_length" type="xs:double"/>
+ <xs:attribute name="id_source" type="phy:id_source"/>
+ </xs:complexType>
+ <!-- Taxonomy:-->
+ <xs:complexType name="Taxonomy">
+ <xs:annotation>
+ <xs:documentation> Element Taxonomy is used to describe taxonomic information for a clade. Element 'code' is
+ intended to store UniProt/Swiss-Prot style organism codes (e.g. 'APLCA' for the California sea hare 'Aplysia
+ californica') or other styles of mnemonics (e.g. 'Aca'). Element 'authority' is used to keep the authority,
+ such as 'J. G. Cooper, 1863', associated with the 'scientific_name'. Element 'id' is used for a unique
+ identifier of a taxon (for example '6500' with 'ncbi_taxonomy' as 'provider' for the California sea hare).
+ Attribute 'id_source' is used to link other elements to a taxonomy (on the xml-level).</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="id" type="phy:Id" minOccurs="0"/>
+ <xs:element name="code" type="phy:TaxonomyCode" minOccurs="0"/>
+ <xs:element name="scientific_name" type="xs:token" minOccurs="0"/>
+ <xs:element name="authority" type="xs:token" minOccurs="0"/>
+ <xs:element name="common_name" type="xs:token" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="synonym" type="xs:token" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="rank" type="phy:Rank" minOccurs="0"/>
+ <xs:element name="uri" type="phy:Uri" minOccurs="0"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="id_source" type="phy:id_source"/>
+ </xs:complexType>
+ <xs:simpleType name="TaxonomyCode">
+ <xs:restriction base="xs:token">
+ <xs:pattern value="[a-zA-Z0-9_]{2,10}"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:simpleType name="Rank">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="domain"/>
+ <xs:enumeration value="superkingdom"/>
+ <xs:enumeration value="kingdom"/>
+ <xs:enumeration value="subkingdom"/>
+ <xs:enumeration value="branch"/>
+ <xs:enumeration value="infrakingdom"/>
+ <xs:enumeration value="superphylum"/>
+ <xs:enumeration value="phylum"/>
+ <xs:enumeration value="subphylum"/>
+ <xs:enumeration value="infraphylum"/>
+ <xs:enumeration value="microphylum"/>
+ <xs:enumeration value="superdivision"/>
+ <xs:enumeration value="division"/>
+ <xs:enumeration value="subdivision"/>
+ <xs:enumeration value="infradivision"/>
+ <xs:enumeration value="superclass"/>
+ <xs:enumeration value="class"/>
+ <xs:enumeration value="subclass"/>
+ <xs:enumeration value="infraclass"/>
+ <xs:enumeration value="superlegion"/>
+ <xs:enumeration value="legion"/>
+ <xs:enumeration value="sublegion"/>
+ <xs:enumeration value="infralegion"/>
+ <xs:enumeration value="supercohort"/>
+ <xs:enumeration value="cohort"/>
+ <xs:enumeration value="subcohort"/>
+ <xs:enumeration value="infracohort"/>
+ <xs:enumeration value="superorder"/>
+ <xs:enumeration value="order"/>
+ <xs:enumeration value="suborder"/>
+ <xs:enumeration value="superfamily"/>
+ <xs:enumeration value="family"/>
+ <xs:enumeration value="subfamily"/>
+ <xs:enumeration value="supertribe"/>
+ <xs:enumeration value="tribe"/>
+ <xs:enumeration value="subtribe"/>
+ <xs:enumeration value="infratribe"/>
+ <xs:enumeration value="genus"/>
+ <xs:enumeration value="subgenus"/>
+ <xs:enumeration value="superspecies"/>
+ <xs:enumeration value="species"/>
+ <xs:enumeration value="subspecies"/>
+ <xs:enumeration value="variety"/>
+ <xs:enumeration value="subvariety"/>
+ <xs:enumeration value="form"/>
+ <xs:enumeration value="subform"/>
+ <xs:enumeration value="cultivar"/>
+ <xs:enumeration value="strain"/>
+ <xs:enumeration value="unknown"/>
+ <xs:enumeration value="other"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!-- Sequence:-->
+ <xs:complexType name="Sequence">
+ <xs:annotation>
+ <xs:documentation> Element Sequence is used to represent a molecular sequence (Protein, DNA, RNA) associated
+ with a node. 'symbol' is a short (maximal ten characters) symbol of the sequence (e.g. 'ACTM') whereas
+ 'name' is used for the full name (e.g. 'muscle Actin'). 'location' is used for the location of a sequence on
+ a genome/chromosome. The actual sequence can be stored with the 'mol_seq' element. Attribute 'type' is used
+ to indicate the type of sequence ('dna', 'rna', or 'protein'). One intended use for 'id_ref' is to link a
+ sequence to a taxonomy (via the taxonomy's 'id_source') in case of multiple sequences and taxonomies per
+ node. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="symbol" type="phy:SequenceSymbol" minOccurs="0"/>
+ <xs:element name="accession" type="phy:Accession" minOccurs="0"/>
+ <xs:element name="name" type="xs:token" minOccurs="0"/>
+ <xs:element name="location" type="xs:token" minOccurs="0"/>
+ <xs:element name="mol_seq" type="phy:MolSeq" minOccurs="0"/>
+ <xs:element name="uri" type="phy:Uri" minOccurs="0"/>
+ <xs:element name="annotation" type="phy:Annotation" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="domain_architecture" type="phy:DomainArchitecture" minOccurs="0"/>
+ <xs:any minOccurs="0" maxOccurs="unbounded" processContents="lax" namespace="##other"/>
+ </xs:sequence>
+ <xs:attribute name="type" type="phy:SequenceType"/>
+ <xs:attribute name="id_source" type="phy:id_source"/>
+ <xs:attribute name="id_ref" type="phy:id_ref"/>
+ </xs:complexType>
+ <xs:simpleType name="SequenceSymbol">
+ <xs:restriction base="xs:token">
+ <xs:pattern value="\S{1,20}"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:complexType name="MolSeq">
+ <xs:annotation>
+ <xs:documentation> Element 'mol_seq' is used to store molecular sequences. The 'is_aligned' attribute is used
+ to indicated that this molecular sequence is aligned with all other sequences in the same phylogeny for
+ which 'is aligned' is true as well (which, in most cases, means that gaps were introduced, and that all
+ sequences for which 'is aligned' is true must have the same length).</xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="is_aligned" type="xs:boolean"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <xs:simpleType name="SequenceType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="rna"/>
+ <xs:enumeration value="dna"/>
+ <xs:enumeration value="protein"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!-- Accession:-->
+ <xs:complexType name="Accession">
+ <xs:annotation>
+ <xs:documentation> Element Accession is used to capture the local part in a sequence identifier (e.g. 'P17304'
+ in 'UniProtKB:P17304', in which case the 'source' attribute would be 'UniProtKB'). </xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="source" type="xs:token" use="required"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- DomainArchitecture: -->
+ <xs:complexType name="DomainArchitecture">
+ <xs:annotation>
+ <xs:documentation> This is used describe the domain architecture of a protein. Attribute 'length' is the total
+ length of the protein</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="domain" type="phy:ProteinDomain" minOccurs="1" maxOccurs="unbounded"/>
+ </xs:sequence>
+ <xs:attribute name="length" type="xs:nonNegativeInteger"/>
+ </xs:complexType>
+ <xs:complexType name="ProteinDomain">
+ <xs:annotation>
+ <xs:documentation> To represent an individual domain in a domain architecture. The name/unique identifier is
+ described via the 'id' attribute. 'confidence' can be used to store (i.e.) E-values.</xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="from" type="xs:nonNegativeInteger" use="required"/>
+ <xs:attribute name="to" type="xs:nonNegativeInteger" use="required"/>
+ <xs:attribute name="confidence" type="xs:double"/>
+ <xs:attribute name="id" type="xs:token"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Events:-->
+ <xs:complexType name="Events">
+ <xs:annotation>
+ <xs:documentation> Events at the root node of a clade (e.g. one gene duplication). </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="type" type="phy:EventType" minOccurs="0"/>
+ <xs:element name="duplications" type="xs:nonNegativeInteger" minOccurs="0"/>
+ <xs:element name="speciations" type="xs:nonNegativeInteger" minOccurs="0"/>
+ <xs:element name="losses" type="xs:nonNegativeInteger" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ </xs:sequence>
+ </xs:complexType>
+ <xs:simpleType name="EventType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="transfer"/>
+ <xs:enumeration value="fusion"/>
+ <xs:enumeration value="speciation_or_duplication"/>
+ <xs:enumeration value="other"/>
+ <xs:enumeration value="mixed"/>
+ <xs:enumeration value="unassigned"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!--BinaryCharacters:-->
+ <xs:complexType name="BinaryCharacters">
+ <xs:annotation>
+ <xs:documentation> The names and/or counts of binary characters present, gained, and lost at the root of a
+ clade. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="gained" type="phy:BinaryCharacterList" minOccurs="0"/>
+ <xs:element name="lost" type="phy:BinaryCharacterList" minOccurs="0"/>
+ <xs:element name="present" type="phy:BinaryCharacterList" minOccurs="0"/>
+ <xs:element name="absent" type="phy:BinaryCharacterList" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="type" type="xs:token"/>
+ <xs:attribute name="gained_count" type="xs:nonNegativeInteger"/>
+ <xs:attribute name="lost_count" type="xs:nonNegativeInteger"/>
+ <xs:attribute name="present_count" type="xs:nonNegativeInteger"/>
+ <xs:attribute name="absent_count" type="xs:nonNegativeInteger"/>
+ </xs:complexType>
+ <xs:complexType name="BinaryCharacterList">
+ <xs:sequence>
+ <xs:element name="bc" type="xs:token" maxOccurs="unbounded"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- Reference:-->
+ <xs:complexType name="Reference">
+ <xs:annotation>
+ <xs:documentation> A literature reference for a clade. It is recommended to use the 'doi' attribute instead of
+ the free text 'desc' element whenever possible. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="doi" type="xs:token"/>
+ </xs:complexType>
+ <!-- Annotation:-->
+ <xs:complexType name="Annotation">
+ <xs:annotation>
+ <xs:documentation> The annotation of a molecular sequence. It is recommended to annotate by using the optional
+ 'ref' attribute (some examples of acceptable values for the ref attribute: 'GO:0008270',
+ 'KEGG:Tetrachloroethene degradation', 'EC:1.1.1.1'). Optional element 'desc' allows for a free text
+ description. Optional element 'confidence' is used to state the type and value of support for a annotation.
+ Similarly, optional attribute 'evidence' is used to describe the evidence for a annotation as free text
+ (e.g. 'experimental'). Optional element 'property' allows for further, typed and referenced annotations from
+ external resources.</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ <xs:element name="property" type="phy:Property" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="uri" type="phy:Uri" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="ref" type="phy:ref"/>
+ <xs:attribute name="source" type="xs:token"/>
+ <xs:attribute name="evidence" type="xs:token"/>
+ <xs:attribute name="type" type="xs:token"/>
+ </xs:complexType>
+ <!-- Property:-->
+ <xs:complexType name="Property" mixed="true">
+ <xs:annotation>
+ <xs:documentation> Property allows for typed and referenced properties from external resources to be attached
+ to 'Phylogeny', 'Clade', and 'Annotation'. The value of a property is its mixed (free text) content.
+ Attribute 'datatype' indicates the type of a property and is limited to xsd-datatypes (e.g. 'xsd:string',
+ 'xsd:boolean', 'xsd:integer', 'xsd:decimal', 'xsd:float', 'xsd:double', 'xsd:date', 'xsd:anyURI'). Attribute
+ 'applies_to' indicates the item to which a property applies to (e.g. 'node' for the parent node of a clade,
+ 'parent_branch' for the parent branch of a clade). Attribute 'id_ref' allows to attached a property
+ specifically to one element (on the xml-level). Optional attribute 'unit' is used to indicate the unit of
+ the property. An example: <property datatype="xsd:integer" ref="NOAA:depth" applies_to="clade"
+ unit="METRIC:m"> 200 </property> </xs:documentation>
+ </xs:annotation>
+ <xs:attribute name="ref" type="phy:ref" use="required"/>
+ <xs:attribute name="unit" type="phy:ref"/>
+ <xs:attribute name="datatype" type="phy:PropertyDataType" use="required"/>
+ <xs:attribute name="applies_to" type="phy:AppliesTo" use="required"/>
+ <xs:attribute name="id_ref" type="phy:id_ref"/>
+ </xs:complexType>
+ <xs:simpleType name="ref">
+ <xs:restriction base="xs:token">
+ <xs:pattern value="[a-zA-Z0-9_]+:[a-zA-Z0-9_\.\-\s]+"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:simpleType name="AppliesTo">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="phylogeny"/>
+ <xs:enumeration value="clade"/>
+ <xs:enumeration value="node"/>
+ <xs:enumeration value="annotation"/>
+ <xs:enumeration value="parent_branch"/>
+ <xs:enumeration value="other"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <xs:simpleType name="PropertyDataType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="xsd:string"/>
+ <xs:enumeration value="xsd:boolean"/>
+ <xs:enumeration value="xsd:decimal"/>
+ <xs:enumeration value="xsd:float"/>
+ <xs:enumeration value="xsd:double"/>
+ <xs:enumeration value="xsd:duration"/>
+ <xs:enumeration value="xsd:dateTime"/>
+ <xs:enumeration value="xsd:time"/>
+ <xs:enumeration value="xsd:date"/>
+ <xs:enumeration value="xsd:gYearMonth"/>
+ <xs:enumeration value="xsd:gYear"/>
+ <xs:enumeration value="xsd:gMonthDay"/>
+ <xs:enumeration value="xsd:gDay"/>
+ <xs:enumeration value="xsd:gMonth"/>
+ <xs:enumeration value="xsd:hexBinary"/>
+ <xs:enumeration value="xsd:base64Binary"/>
+ <xs:enumeration value="xsd:anyURI"/>
+ <xs:enumeration value="xsd:normalizedString"/>
+ <xs:enumeration value="xsd:token"/>
+ <xs:enumeration value="xsd:integer"/>
+ <xs:enumeration value="xsd:nonPositiveInteger"/>
+ <xs:enumeration value="xsd:negativeInteger"/>
+ <xs:enumeration value="xsd:long"/>
+ <xs:enumeration value="xsd:int"/>
+ <xs:enumeration value="xsd:short"/>
+ <xs:enumeration value="xsd:byte"/>
+ <xs:enumeration value="xsd:nonNegativeInteger"/>
+ <xs:enumeration value="xsd:unsignedLong"/>
+ <xs:enumeration value="xsd:unsignedInt"/>
+ <xs:enumeration value="xsd:unsignedShort"/>
+ <xs:enumeration value="xsd:unsignedByte"/>
+ <xs:enumeration value="xsd:positiveInteger"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!--Uri-->
+ <xs:complexType name="Uri">
+ <xs:annotation>
+ <xs:documentation> A uniform resource identifier. In general, this is expected to be an URL (for example, to
+ link to an image on a website, in which case the 'type' attribute might be 'image' and 'desc' might be
+ 'image of a California sea hare'). </xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:anyURI">
+ <xs:attribute name="desc" type="xs:token"/>
+ <xs:attribute name="type" type="xs:token"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Confidence:-->
+ <xs:complexType name="Confidence">
+ <xs:annotation>
+ <xs:documentation> A general purpose confidence element. For example this can be used to express the bootstrap
+ support value of a clade (in which case the 'type' attribute is 'bootstrap').</xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:double">
+ <xs:attribute name="type" type="xs:token" use="required"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Identifier:-->
+ <xs:complexType name="Id">
+ <xs:annotation>
+ <xs:documentation> A general purpose identifier element. Allows to indicate the provider (or authority) of an
+ identifier. </xs:documentation>
+ </xs:annotation>
+ <xs:simpleContent>
+ <xs:extension base="xs:token">
+ <xs:attribute name="provider" type="xs:token"/>
+ </xs:extension>
+ </xs:simpleContent>
+ </xs:complexType>
+ <!-- Distribution:-->
+ <xs:complexType name="Distribution">
+ <xs:annotation>
+ <xs:documentation> The geographic distribution of the items of a clade (species, sequences), intended for
+ phylogeographic applications. The location can be described either by free text in the 'desc' element and/or
+ by the coordinates of one or more 'Points' (similar to the 'Point' element in Google's KML format) or by
+ 'Polygons'. </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ <xs:element name="point" type="phy:Point" minOccurs="0" maxOccurs="unbounded"/>
+ <xs:element name="polygon" type="phy:Polygon" minOccurs="0" maxOccurs="unbounded"/>
+ </xs:sequence>
+ </xs:complexType>
+ <xs:complexType name="Point">
+ <xs:annotation>
+ <xs:documentation> The coordinates of a point with an optional altitude (used by element 'Distribution').
+ Required attributes are the 'geodetic_datum' used to indicate the geodetic datum (also called 'map datum',
+ for example Google's KML uses 'WGS84'). Attribute 'alt_unit' is the unit for the altitude (e.g. 'meter').
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="lat" type="xs:decimal"/>
+ <xs:element name="long" type="xs:decimal"/>
+ <xs:element name="alt" type="xs:decimal" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="geodetic_datum" type="xs:token" use="required"/>
+ <xs:attribute name="alt_unit" type="xs:token"/>
+ </xs:complexType>
+ <xs:complexType name="Polygon">
+ <xs:annotation>
+ <xs:documentation> A polygon defined by a list of 'Points' (used by element 'Distribution').
+ </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="point" type="phy:Point" minOccurs="3" maxOccurs="unbounded"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- Date:-->
+ <xs:complexType name="Date">
+ <xs:annotation>
+ <xs:documentation> A date associated with a clade/node. Its value can be numerical by using the 'value' element
+ and/or free text with the 'desc' element' (e.g. 'Silurian'). If a numerical value is used, it is recommended
+ to employ the 'unit' attribute to indicate the type of the numerical value (e.g. 'mya' for 'million years
+ ago'). The elements 'minimum' and 'maximum' are used the indicate a range/confidence
+ interval</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="desc" type="xs:token" minOccurs="0"/>
+ <xs:element name="value" type="xs:decimal" minOccurs="0"/>
+ <xs:element name="minimum" type="xs:decimal" minOccurs="0"/>
+ <xs:element name="maximum" type="xs:decimal" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="unit" type="xs:token"/>
+ </xs:complexType>
+ <!-- BranchColor:-->
+ <xs:complexType name="BranchColor">
+ <xs:annotation>
+ <xs:documentation> This indicates the color of a clade when rendered (the color applies to the whole clade
+ unless overwritten by the color(s) of sub clades).</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="red" type="xs:unsignedByte"/>
+ <xs:element name="green" type="xs:unsignedByte"/>
+ <xs:element name="blue" type="xs:unsignedByte"/>
+ </xs:sequence>
+ </xs:complexType>
+ <!-- SequenceRelation:-->
+ <xs:complexType name="SequenceRelation">
+ <xs:annotation>
+ <xs:documentation> This is used to express a typed relationship between two sequences. For example it could be
+ used to describe an orthology (in which case attribute 'type' is 'orthology'). </xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>
+ <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>
+ <xs:attribute name="distance" type="xs:double"/>
+ <xs:attribute name="type" type="phy:SequenceRelationType" use="required"/>
+ </xs:complexType>
+ <xs:simpleType name="SequenceRelationType">
+ <xs:restriction base="xs:token">
+ <xs:enumeration value="orthology"/>
+ <xs:enumeration value="one_to_one_orthology"/>
+ <xs:enumeration value="super_orthology"/>
+ <xs:enumeration value="paralogy"/>
+ <xs:enumeration value="ultra_paralogy"/>
+ <xs:enumeration value="xenology"/>
+ <xs:enumeration value="unknown"/>
+ <xs:enumeration value="other"/>
+ </xs:restriction>
+ </xs:simpleType>
+ <!-- CladeRelation:-->
+ <xs:complexType name="CladeRelation">
+ <xs:annotation>
+ <xs:documentation> This is used to express a typed relationship between two clades. For example it could be
+ used to describe multiple parents of a clade.</xs:documentation>
+ </xs:annotation>
+ <xs:sequence>
+ <xs:element name="confidence" type="phy:Confidence" minOccurs="0"/>
+ </xs:sequence>
+ <xs:attribute name="id_ref_0" type="phy:id_ref" use="required"/>
+ <xs:attribute name="id_ref_1" type="phy:id_ref" use="required"/>
+ <xs:attribute name="distance" type="xs:double"/>
+ <xs:attribute name="type" type="xs:token" use="required"/>
+ </xs:complexType>
+ <!-- Used to link elements together on the xml level:-->
+ <xs:simpleType name="id_source">
+ <xs:restriction base="xs:ID"/>
+ </xs:simpleType>
+ <xs:simpleType name="id_ref">
+ <xs:restriction base="xs:IDREF"/>
+ </xs:simpleType>
+</xs:schema>
}
final SortedSet<String> all_chars = getAllExternalPresentAndGainedCharacters( phy.getRoot() );
final SortedSet<String> human = getAllExternalPresentAndGainedCharacters( phy.getNode( "HUMAN" ) );
- final SortedSet<String> primates = getAllExternalPresentAndGainedCharacters( find( "Primates", phy ) );
- final SortedSet<String> mammalia = getAllExternalPresentAndGainedCharacters( find( "Mammalia", phy ) );
- final SortedSet<String> metazoa = getAllExternalPresentAndGainedCharacters( find( "Metazoa", phy ) );
- final SortedSet<String> fungi = getAllExternalPresentAndGainedCharacters( find( "Fungi", phy ) );
- final SortedSet<String> plants = getAllExternalPresentAndGainedCharacters( find( "Viridiplantae", phy ) );
+ final SortedSet<String> primates = getAllExternalPresentAndGainedCharacters( phy.getNode( find( "Primates", phy ) ) );
+ final SortedSet<String> mammalia = getAllExternalPresentAndGainedCharacters( phy.getNode(find( "Mammalia", phy ) ) );
+ final SortedSet<String> metazoa = getAllExternalPresentAndGainedCharacters( phy.getNode(find( "Metazoa", phy ) ) );
+ final SortedSet<String> fungi = getAllExternalPresentAndGainedCharacters( phy.getNode(find( "Fungi", phy ) ) );
+ final SortedSet<String> plants = getAllExternalPresentAndGainedCharacters( phy.getNode(find( "Viridiplantae", phy ) ) );
System.out.println( "Sum of all external characters:\t" + all_chars.size() );
System.out.println();
final List<PhylogenyNode> ext = phy.getRoot().getAllExternalDescendants();
}
}
- private static PhylogenyNode find( final String s, final Phylogeny phy ) {
- final List<PhylogenyNode> l = PhylogenyMethods.searchData( s, phy, true, false, false, false, 0 );
+ private static Long find( final String s, final Phylogeny phy ) {
+ final List<Long> l = PhylogenyMethods.searchData( s, phy, true, false, false, false, 0 );
if ( l.size() != 1 ) {
System.err.println( "error: " + s );
System.exit( -1 );
git://source.jalview.org / jalview.git / commitdiff