X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fio%2FEmblFlatFile.java;h=bfae4ed37eb64a446967420e9bfd40fbe8315aa9;hb=cb79cecb722e9e3b17eb3013ed24f61c5fc6c800;hp=5be4364d60ea1574841ca559d35af2beb2df988e;hpb=d4ed63b811b2b57149af11ba0ccbe11c6845acba;p=jalview.git diff --git a/src/jalview/io/EmblFlatFile.java b/src/jalview/io/EmblFlatFile.java index 5be4364..bfae4ed 100644 --- a/src/jalview/io/EmblFlatFile.java +++ b/src/jalview/io/EmblFlatFile.java @@ -3,14 +3,17 @@ package jalview.io; import java.io.IOException; import java.text.ParseException; import java.util.ArrayList; +import java.util.Arrays; import java.util.HashMap; import java.util.Hashtable; import java.util.List; import java.util.Map; import java.util.Map.Entry; +import java.util.TreeMap; import jalview.bin.Cache; import jalview.datamodel.DBRefEntry; +import jalview.datamodel.DBRefSource; import jalview.datamodel.FeatureProperties; import jalview.datamodel.Mapping; import jalview.datamodel.Sequence; @@ -18,6 +21,7 @@ import jalview.datamodel.SequenceFeature; import jalview.datamodel.SequenceI; import jalview.util.DBRefUtils; import jalview.util.DnaUtils; +import jalview.util.MapList; import jalview.util.MappingUtils; /** @@ -41,6 +45,10 @@ import jalview.util.MappingUtils; */ public class EmblFlatFile extends AlignFile // FileParse { + private static final String QUOTE = "\""; + + private static final String DOUBLED_QUOTE = QUOTE + QUOTE; + /** * A data bean class to hold values parsed from one CDS Feature (FT) */ @@ -56,6 +64,8 @@ public class EmblFlatFile extends AlignFile // FileParse String proteinId; // from CDS /protein_id + List xrefs = new ArrayList<>(); // from CDS /db_xref qualifiers + Map cdsProps = new Hashtable<>(); // CDS other qualifiers } @@ -74,11 +84,14 @@ public class EmblFlatFile extends AlignFile // FileParse private int length = 128; // from ID (7th token), with usable default - private List dbrefs; // from DR and also CDS /db_xref qualifiers + private List dbrefs; // from DR private String sequenceString; // from SQ lines - private List cds; + /* + * parsed CDS data fields, keyed by protein_id + */ + private Map cds; /** * Constructor @@ -92,7 +105,11 @@ public class EmblFlatFile extends AlignFile // FileParse super(false, fp); // don't parse immediately this.sourceDb = sourceId; dbrefs = new ArrayList<>(); - cds = new ArrayList<>(); + + /* + * using TreeMap gives CDS sequences in alphabetical, so readable, order + */ + cds = new TreeMap<>(String.CASE_INSENSITIVE_ORDER); } /** @@ -101,6 +118,7 @@ public class EmblFlatFile extends AlignFile // FileParse * * @throws IOException */ + @Override public void parse() throws IOException { String line = nextLine(); @@ -131,7 +149,7 @@ public class EmblFlatFile extends AlignFile // FileParse line = nextLine(); } } - assembleSequence(); + buildSequence(); } /** @@ -305,10 +323,14 @@ public class EmblFlatFile extends AlignFile // FileParse return nextLine(); } + /* + * parse location - which may be over more than one line e.g. EAW51554 + */ CdsData data = new CdsData(); - data.cdsLocation = tokens[2]; + StringBuilder sb = new StringBuilder().append(tokens[2]); + line = parseFeatureQualifier(sb, "CDS"); + data.cdsLocation = sb.toString(); - line = nextLine(); while (line != null) { if (!line.startsWith("FT ")) // 4 spaces @@ -319,93 +341,139 @@ public class EmblFlatFile extends AlignFile // FileParse /* * extract qualifier, e.g. FT /protein_id="CAA37824.1" + * - the value may extend over more than one line + * - if the value has enclosing quotes, these are removed + * - escaped double quotes ("") are reduced to a single character */ int slashPos = line.indexOf('/'); if (slashPos == -1) { Cache.log.error("Unexpected EMBL line ignored: " + line); + line = nextLine(); continue; } int eqPos = line.indexOf('=', slashPos + 1); if (eqPos == -1) { - Cache.log.error("Unexpected EMBL line ignored: " + line); + // can happen, e.g. /ribosomal_slippage + // Cache.log.error("Unexpected EMBL line ignored: " + line); + line = nextLine(); continue; } String qualifier = line.substring(slashPos + 1, eqPos); String value = line.substring(eqPos + 1); - if (value.startsWith("\"") && value.endsWith("\"")) - { - value = value.substring(1, value.length() - 1); - } + value = removeQuotes(value); + sb = new StringBuilder().append(value); + line = parseFeatureQualifier(sb, qualifier); + String featureValue = sb.toString(); if ("protein_id".equals(qualifier)) { - data.proteinId = value; - line = nextLine(); + data.proteinId = featureValue; } else if ("codon_start".equals(qualifier)) { try { - data.codonStart = Integer.parseInt(value.trim()); + data.codonStart = Integer.parseInt(featureValue.trim()); } catch (NumberFormatException e) { Cache.log.error("Invalid codon_start in XML for " + this.accession + ": " + e.getMessage()); } - line = nextLine(); } else if ("db_xref".equals(qualifier)) { - String[] parts = value.split(":"); + String[] parts = featureValue.split(":"); if (parts.length == 2) { String db = parts[0].trim(); db = DBRefUtils.getCanonicalName(db); DBRefEntry dbref = new DBRefEntry(db, "0", parts[1].trim()); - this.dbrefs.add(dbref); + data.xrefs.add(dbref); } - line = nextLine(); } else if ("product".equals(qualifier)) { - // sometimes name is returned e.g. for V00488 - data.proteinName = value; - line = nextLine(); + data.proteinName = featureValue; } else if ("translation".equals(qualifier)) { - line = readTranslation(value, data); + data.translation = featureValue; } - else if (!"".equals(value)) + else if (!"".equals(featureValue)) { // throw anything else into the additional properties hash - data.cdsProps.put(qualifier, value); - line = nextLine(); + data.cdsProps.put(qualifier, featureValue); } } - this.cds.add(data); + if (data.proteinId != null) + { + this.cds.put(data.proteinId, data); + } + else + { + Cache.log.error("Ignoring CDS feature with no protein_id for " + + sourceDb + ":" + accession); + } return line; } /** - * Reads and returns the CDS translation from one or more lines of the file, - * and returns the next line after that + * Removes leading or trailing double quotes (") unless doubled, and changes + * any 'escaped' (doubled) double quotes to single characters. As per the + * Feature Table specification for Qualifiers, Free Text. * * @param value - * the first line of the translation (likely quoted) - * @param data * @return - * @throws IOException */ - String readTranslation(String value, CdsData data) throws IOException + static String removeQuotes(String value) { - StringBuilder sb = new StringBuilder(this.length / 3 + 1); - sb.append(value.replace("\"", "")); + if (value == null) + { + return null; + } + if (value.startsWith(QUOTE) && !value.startsWith(DOUBLED_QUOTE)) + { + value = value.substring(1); + } + if (value.endsWith(QUOTE) && !value.endsWith(DOUBLED_QUOTE)) + { + value = value.substring(0, value.length() - 1); + } + value = value.replace(DOUBLED_QUOTE, QUOTE); + return value; + } + /** + * Reads the value of a feature (FT) qualifier from one or more lines of the + * file, and returns the next line after that. Values are appended to the + * string buffer, which should be already primed with the value read from the + * first line for the qualifier (with any leading double quote removed). + * Enclosing double quotes are removed, and escaped (repeated) double quotes + * reduced to one only. For example for + * + * 
+   * FT      /note="gene_id=hCG28070.3 
+   * FT      ""foobar"" isoform=CRA_b"
+   * the returned value is
+   * gene_id=hCG28070.3 "foobar" isoform=CRA_b
+   *
+ * + * Note the side-effect of this method, to advance data reading to the next + * line after the feature qualifier. + * + * @param sb + * a string buffer primed with the first line of the value + * @param qualifierName + * @return + * @throws IOException + */ + String parseFeatureQualifier(StringBuilder sb, String qualifierName) + throws IOException + { String line; while ((line = nextLine()) != null) { @@ -416,17 +484,31 @@ public class EmblFlatFile extends AlignFile // FileParse String[] tokens = line.split(WHITESPACE); if (tokens.length < 2) { - Cache.log.error("Ignoring bad EMBL line: " + line); + Cache.log.error("Ignoring bad EMBL line for " + this.accession + + ": " + line); break; } if (tokens[1].startsWith("/")) { break; // next feature qualifier } - sb.append(tokens[1].replace("\"", "")); - } - data.translation = sb.toString(); + /* + * heuristic rule: most multi-line value (e.g. /product) are text, + * so add a space for word boundary at a new line; not for translation + */ + if (!"translation".equals(qualifierName) + && !"CDS".equals(qualifierName)) + { + sb.append(" "); + } + + /* + * remove trailing " and unescape doubled "" + */ + String data = removeQuotes(tokens[1]); + sb.append(data); + } return line; } @@ -434,8 +516,14 @@ public class EmblFlatFile extends AlignFile // FileParse /** * Constructs and saves the sequence from parsed components */ - void assembleSequence() + void buildSequence() { + if (this.accession == null || this.sequenceString == null) + { + Cache.log.error("Failed to parse data from EMBL"); + return; + } + String name = this.accession; if (this.sourceDb != null) { @@ -457,7 +545,7 @@ public class EmblFlatFile extends AlignFile // FileParse seq.addDBRef(dbref); } - processAllCDS(seq); + processCDSFeatures(seq); seq.deriveSequence(); @@ -470,25 +558,25 @@ public class EmblFlatFile extends AlignFile // FileParse * * @param seq */ - protected void processAllCDS(SequenceI seq) + protected void processCDSFeatures(SequenceI seq) { /* * record protein products found to avoid duplication i.e. >1 CDS with * the same /protein_id [though not sure I can find an example of this] */ Map proteins = new HashMap<>(); - for (CdsData data : cds) + for (CdsData data : cds.values()) { - processOneCDS(seq, data, proteins); + processCDSFeature(seq, data, proteins); } } /** - * Processes the parsed CDS feature data to + * Processes data for one parsed CDS feature to *
    - *
  • add a CDS feature to the sequence for each CDS start-end range
    • *
  • create a protein product sequence for the translation
    • *
  • create a cross-reference to protein with mapping from dna
    • + *
  • add a CDS feature to the sequence for each CDS start-end range
    • *
  • add any CDS dbrefs to the sequence and to the protein product
    • *
* @@ -497,13 +585,16 @@ public class EmblFlatFile extends AlignFile // FileParse * @param proteins * map of protein products so far derived from CDS data */ - void processOneCDS(SequenceI dna, CdsData data, + void processCDSFeature(SequenceI dna, CdsData data, Map proteins) { /* * parse location into a list of [start, end, start, end] positions */ int[] exons = getCdsRanges(this.accession, data.cdsLocation); + + MapList maplist = buildMappingToProtein(dna, exons, data); + int exonNumber = 0; for (int xint = 0; exons != null && xint < exons.length - 1; xint += 2) @@ -531,9 +622,128 @@ public class EmblFlatFile extends AlignFile // FileParse sf.setValue(FeatureProperties.EXONPRODUCT, data.proteinName); dna.addSequenceFeature(sf); + } + + boolean hasUniprotDbref = false; + for (DBRefEntry xref : data.xrefs) + { + dna.addDBRef(xref); + if (xref.getSource().equals(DBRefSource.UNIPROT)) + { + /* + * construct (or find) the sequence for (data.protein_id, data.translation) + */ + SequenceI protein = buildProteinProduct(dna, xref, data, proteins); + Mapping map = new Mapping(protein, maplist); + map.setMappedFromId(data.proteinId); + xref.setMap(map); + + /* + * add DBRefs with mappings from dna to protein and the inverse + */ + DBRefEntry db1 = new DBRefEntry(sourceDb, version, accession); + db1.setMap(new Mapping(dna, maplist.getInverse())); + protein.addDBRef(db1); + + hasUniprotDbref = true; + } + } + + /* + * if we have a product (translation) but no explicit Uniprot dbref + * (example: EMBL M19487 protein_id AAB02592.1) + * then construct mappings to an assumed EMBLCDSPROTEIN accession + */ + if (!hasUniprotDbref) + { + SequenceI protein = proteins.get(data.proteinId); + if (protein == null) + { + protein = new Sequence(data.proteinId, data.translation); + protein.setDescription(data.proteinName); + proteins.put(data.proteinId, protein); + } + // assuming CDSPROTEIN sequence version = dna version (?!) + DBRefEntry db1 = new DBRefEntry(DBRefSource.EMBLCDSProduct, + this.version, data.proteinId); + protein.addDBRef(db1); + + DBRefEntry dnaToEmblProteinRef = new DBRefEntry( + DBRefSource.EMBLCDSProduct, this.version, data.proteinId); + Mapping map = new Mapping(protein, maplist); + map.setMappedFromId(data.proteinId); + dnaToEmblProteinRef.setMap(map); + dna.addDBRef(dnaToEmblProteinRef); + } - linkProteinProduct(dna, data, proteins); + /* + * comment brought forward from EmblXmlSource, lines 447-451: + * TODO: if retrieved from EMBLCDS, add a DBRef back to the parent EMBL + * sequence with the exon map; if given a dataset reference, search + * dataset for parent EMBL sequence if it exists and set its map; + * make a new feature annotating the coding contig + */ + } + + /** + * Computes a mapping from CDS positions in DNA sequence to protein product + * positions, with allowance for stop codon or incomplete start codon + * + * @param dna + * @param exons + * @param data + * @return + */ + MapList buildMappingToProtein(final SequenceI dna, final int[] exons, + final CdsData data) + { + MapList dnaToProteinMapping = null; + int peptideLength = data.translation.length(); + + int[] proteinRange = new int[] { 1, peptideLength }; + if (exons != null && exons.length > 0) + { + /* + * We were able to parse 'location'; do a final + * product length truncation check + */ + int[] cdsRanges = adjustForProteinLength(peptideLength, exons); + dnaToProteinMapping = new MapList(cdsRanges, proteinRange, 3, 1); + } + else + { + /* + * workaround until we handle all 'location' formats fully + * e.g. X53828.1:60..1058 or <123..>289 + */ + Cache.log.error(String.format( + "Implementation Notice: EMBLCDS location '%s'not properly supported yet" + + " - Making up the CDNA region of (%s:%s)... may be incorrect", + data.cdsLocation, sourceDb, this.accession)); + + int completeCodonsLength = 1 - data.codonStart + dna.getLength(); + int mappedDnaEnd = dna.getEnd(); + if (peptideLength * 3 == completeCodonsLength) + { + // this might occur for CDS sequences where no features are marked + Cache.log.warn("Assuming no stop codon at end of cDNA fragment"); + mappedDnaEnd = dna.getEnd(); + } + else if ((peptideLength + 1) * 3 == completeCodonsLength) + { + Cache.log.warn("Assuming stop codon at end of cDNA fragment"); + mappedDnaEnd = dna.getEnd() - 3; + } + + if (mappedDnaEnd != -1) + { + int[] cdsRanges = new int[] { + dna.getStart() + (data.codonStart - 1), mappedDnaEnd }; + dnaToProteinMapping = new MapList(cdsRanges, proteinRange, 3, 1); + } } + + return dnaToProteinMapping; } /** @@ -541,31 +751,37 @@ public class EmblFlatFile extends AlignFile // FileParse * one), and dbrefs with mappings from CDS to protein and the reverse * * @param dna + * @param xref * @param data * @param proteins + * @return */ - void linkProteinProduct(SequenceI dna, CdsData data, Map proteins) + SequenceI buildProteinProduct(SequenceI dna, DBRefEntry xref, + CdsData data, Map proteins) { /* * check we have some data to work with */ if (data.proteinId == null || data.translation == null) { - return; + return null; } - + /* * Construct the protein sequence (if not already seen) */ - SequenceI protein = proteins.get(data.proteinId); + String proteinSeqName = xref.getSource() + "|" + xref.getAccessionId(); + SequenceI protein = proteins.get(proteinSeqName); if (protein == null) { - protein = new Sequence(data.proteinId, data.translation, 1, + protein = new Sequence(proteinSeqName, data.translation, 1, data.translation.length()); protein.setDescription(data.proteinName != null ? data.proteinName : "Protein Product from " + sourceDb); - proteins.put(data.proteinId, protein); + proteins.put(proteinSeqName, protein); } + + return protein; } /** @@ -604,4 +820,81 @@ public class EmblFlatFile extends AlignFile // FileParse { return null; } + + /** + * Truncates (if necessary) the exon intervals to match 3 times the length of + * the protein; also accepts 3 bases longer (for stop codon not included in + * protein) + * + * @param proteinLength + * @param exon + * an array of [start, end, start, end...] intervals + * @return the same array (if unchanged) or a truncated copy + */ + static int[] adjustForProteinLength(int proteinLength, int[] exon) + { + if (proteinLength <= 0 || exon == null) + { + return exon; + } + int expectedCdsLength = proteinLength * 3; + int exonLength = MappingUtils.getLength(Arrays.asList(exon)); + + /* + * if exon length matches protein, or is shorter, or longer by the + * length of a stop codon (3 bases), then leave it unchanged + */ + if (expectedCdsLength >= exonLength + || expectedCdsLength == exonLength - 3) + { + return exon; + } + + int origxon[]; + int sxpos = -1; + int endxon = 0; + origxon = new int[exon.length]; + System.arraycopy(exon, 0, origxon, 0, exon.length); + int cdspos = 0; + for (int x = 0; x < exon.length; x += 2) + { + cdspos += Math.abs(exon[x + 1] - exon[x]) + 1; + if (expectedCdsLength <= cdspos) + { + // advanced beyond last codon. + sxpos = x; + if (expectedCdsLength != cdspos) + { + // System.err + // .println("Truncating final exon interval on region by " + // + (cdspos - cdslength)); + } + + /* + * shrink the final exon - reduce end position if forward + * strand, increase it if reverse + */ + if (exon[x + 1] >= exon[x]) + { + endxon = exon[x + 1] - cdspos + expectedCdsLength; + } + else + { + endxon = exon[x + 1] + cdspos - expectedCdsLength; + } + break; + } + } + + if (sxpos != -1) + { + // and trim the exon interval set if necessary + int[] nxon = new int[sxpos + 2]; + System.arraycopy(exon, 0, nxon, 0, sxpos + 2); + nxon[sxpos + 1] = endxon; // update the end boundary for the new exon + // set + exon = nxon; + } + return exon; + } }