*/
package jalview.ws.sifts;
-import jalview.analysis.AlignSeq;
-import jalview.api.DBRefEntryI;
-import jalview.api.SiftsClientI;
-import jalview.datamodel.DBRefEntry;
-import jalview.datamodel.DBRefSource;
-import jalview.datamodel.SequenceI;
-import jalview.io.StructureFile;
-import jalview.schemes.ResidueProperties;
-import jalview.structure.StructureMapping;
-import jalview.util.Comparison;
-import jalview.util.DBRefUtils;
-import jalview.util.Format;
-import jalview.xml.binding.sifts.Entry;
-import jalview.xml.binding.sifts.Entry.Entity;
-import jalview.xml.binding.sifts.Entry.Entity.Segment;
-import jalview.xml.binding.sifts.Entry.Entity.Segment.ListMapRegion.MapRegion;
-import jalview.xml.binding.sifts.Entry.Entity.Segment.ListResidue.Residue;
-import jalview.xml.binding.sifts.Entry.Entity.Segment.ListResidue.Residue.CrossRefDb;
-import jalview.xml.binding.sifts.Entry.Entity.Segment.ListResidue.Residue.ResidueDetail;
-
import java.io.File;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.util.zip.GZIPInputStream;
import javax.xml.bind.JAXBContext;
+import javax.xml.bind.JAXBElement;
import javax.xml.bind.Unmarshaller;
import javax.xml.stream.XMLInputFactory;
import javax.xml.stream.XMLStreamReader;
-import MCview.Atom;
-import MCview.PDBChain;
+import jalview.analysis.AlignSeq;
+import jalview.analysis.scoremodels.ScoreMatrix;
+import jalview.analysis.scoremodels.ScoreModels;
+import jalview.api.DBRefEntryI;
+import jalview.api.SiftsClientI;
+import jalview.datamodel.DBRefEntry;
+import jalview.datamodel.DBRefSource;
+import jalview.datamodel.SequenceI;
+import jalview.io.BackupFiles;
+import jalview.io.StructureFile;
+import jalview.schemes.ResidueProperties;
+import jalview.structure.StructureMapping;
+import jalview.util.Comparison;
+import jalview.util.DBRefUtils;
+import jalview.util.Format;
+import jalview.util.Platform;
+import jalview.xml.binding.sifts.Entry;
+import jalview.xml.binding.sifts.Entry.Entity;
+import jalview.xml.binding.sifts.Entry.Entity.Segment;
+import jalview.xml.binding.sifts.Entry.Entity.Segment.ListMapRegion.MapRegion;
+import jalview.xml.binding.sifts.Entry.Entity.Segment.ListResidue.Residue;
+import jalview.xml.binding.sifts.Entry.Entity.Segment.ListResidue.Residue.CrossRefDb;
+import jalview.xml.binding.sifts.Entry.Entity.Segment.ListResidue.Residue.ResidueDetail;
+import mc_view.Atom;
+import mc_view.PDBChain;
public class SiftsClient implements SiftsClientI
{
private CoordinateSys seqCoordSys = CoordinateSys.UNIPROT;
+ /**
+ * PDB sequence position to sequence coordinate mapping as derived from SIFTS
+ * record for the identified SeqCoordSys Used for lift-over from sequence
+ * derived from PDB (with first extracted PDBRESNUM as 'start' to the sequence
+ * being annotated with PDB data
+ */
+ private jalview.datamodel.Mapping seqFromPdbMapping;
+
private static final int BUFFER_SIZE = 4096;
- public static final int UNASSIGNED = -1;
+ public static final int UNASSIGNED = Integer.MIN_VALUE;
private static final int PDB_RES_POS = 0;
private static final int PDB_ATOM_POS = 1;
+ private static final int PDBE_POS = 2;
+
private static final String NOT_OBSERVED = "Not_Observed";
private static final String SIFTS_FTP_BASE_URL = "http://ftp.ebi.ac.uk/pub/databases/msd/sifts/xml/";
private enum CoordinateSys
{
UNIPROT("UniProt"), PDB("PDBresnum"), PDBe("PDBe");
+
private String name;
private CoordinateSys(String name)
private enum ResidueDetailType
{
- NAME_SEC_STRUCTURE("nameSecondaryStructure"), CODE_SEC_STRUCTURE(
- "codeSecondaryStructure"), ANNOTATION("Annotation");
+ NAME_SEC_STRUCTURE("nameSecondaryStructure"),
+ CODE_SEC_STRUCTURE("codeSecondaryStructure"), ANNOTATION("Annotation");
+
private String code;
private ResidueDetailType(String code)
XMLStreamReader streamReader = XMLInputFactory.newInstance()
.createXMLStreamReader(gzis);
Unmarshaller um = jc.createUnmarshaller();
- return (Entry) um.unmarshal(streamReader);
+ JAXBElement<Entry> jbe = um.unmarshal(streamReader, Entry.class);
+ return jbe.getValue();
} catch (Exception e)
{
e.printStackTrace();
SiftsSettings.getCacheThresholdInDays()))
{
File oldSiftsFile = new File(siftsFileName + "_old");
- siftsFile.renameTo(oldSiftsFile);
+ BackupFiles.moveFileToFile(siftsFile, oldSiftsFile);
try
{
siftsFile = downloadSiftsFile(pdbId.toLowerCase());
} catch (IOException e)
{
e.printStackTrace();
- oldSiftsFile.renameTo(siftsFile);
+ BackupFiles.moveFileToFile(oldSiftsFile, siftsFile);
return new File(siftsFileName);
}
}
+ else
+ {
+ return siftsFile;
+ }
}
try
{
try
{
attr = Files.readAttributes(filePath, BasicFileAttributes.class);
- diffInDays = (int) ((new Date().getTime() - attr.lastModifiedTime()
- .toMillis()) / (1000 * 60 * 60 * 24));
+ diffInDays = (int) ((new Date().getTime()
+ - attr.lastModifiedTime().toMillis())
+ / (1000 * 60 * 60 * 24));
// System.out.println("Diff in days : " + diffInDays);
} catch (IOException e)
{
* @throws SiftsException
* @throws IOException
*/
- public static File downloadSiftsFile(String pdbId) throws SiftsException,
- IOException
+ public static File downloadSiftsFile(String pdbId)
+ throws SiftsException, IOException
{
if (pdbId.contains(".cif"))
{
}
String siftFile = pdbId + ".xml.gz";
String siftsFileFTPURL = SIFTS_FTP_BASE_URL + siftFile;
- String downloadedSiftsFile = SiftsSettings.getSiftDownloadDirectory()
- + siftFile;
- File siftsDownloadDir = new File(
- SiftsSettings.getSiftDownloadDirectory());
- if (!siftsDownloadDir.exists())
+
+ /*
+ * Download the file from URL to either
+ * Java: directory of cached downloaded SIFTS files
+ * Javascript: temporary 'file' (in-memory cache)
+ */
+ File downloadTo = null;
+ if (Platform.isJS())
+ {
+ downloadTo = File.createTempFile(siftFile, ".xml.gz");
+ }
+ else
{
- siftsDownloadDir.mkdirs();
+ downloadTo = new File(
+ SiftsSettings.getSiftDownloadDirectory() + siftFile);
+ File siftsDownloadDir = new File(
+ SiftsSettings.getSiftDownloadDirectory());
+ if (!siftsDownloadDir.exists())
+ {
+ siftsDownloadDir.mkdirs();
+ }
}
+
// System.out.println(">> Download ftp url : " + siftsFileFTPURL);
+ // long now = System.currentTimeMillis();
URL url = new URL(siftsFileFTPURL);
URLConnection conn = url.openConnection();
InputStream inputStream = conn.getInputStream();
- FileOutputStream outputStream = new FileOutputStream(
- downloadedSiftsFile);
+ FileOutputStream outputStream = new FileOutputStream(downloadTo);
byte[] buffer = new byte[BUFFER_SIZE];
int bytesRead = -1;
while ((bytesRead = inputStream.read(buffer)) != -1)
}
outputStream.close();
inputStream.close();
- // System.out.println(">>> File downloaded : " + downloadedSiftsFile);
- return new File(downloadedSiftsFile);
+ // System.out.println(">>> File downloaded : " + downloadedSiftsFile
+ // + " took " + (System.currentTimeMillis() - now) + "ms");
+ return downloadTo;
}
/**
{
continue;
}
- String canonicalSource = DBRefUtils.getCanonicalName(dbRef
- .getSource());
+ String canonicalSource = DBRefUtils
+ .getCanonicalName(dbRef.getSource());
if (isValidDBRefEntry(dbRef)
- && (canonicalSource.equalsIgnoreCase(DBRefSource.UNIPROT) || canonicalSource
- .equalsIgnoreCase(DBRefSource.PDB)))
+ && (canonicalSource.equalsIgnoreCase(DBRefSource.UNIPROT)
+ || canonicalSource.equalsIgnoreCase(DBRefSource.PDB)))
{
return dbRef;
}
public StructureMapping getSiftsStructureMapping(SequenceI seq,
String pdbFile, String chain) throws SiftsException
{
+ SequenceI aseq = seq;
+ while (seq.getDatasetSequence() != null)
+ {
+ seq = seq.getDatasetSequence();
+ }
structId = (chain == null) ? pdbId : pdbId + "|" + chain;
System.out.println("Getting SIFTS mapping for " + structId + ": seq "
+ seq.getName());
HashMap<Integer, int[]> mapping = getGreedyMapping(chain, seq, ps);
String mappingOutput = mappingDetails.toString();
- StructureMapping siftsMapping = new StructureMapping(seq, pdbFile,
- pdbId, chain, mapping, mappingOutput);
+ StructureMapping siftsMapping = new StructureMapping(aseq, pdbFile,
+ pdbId, chain, mapping, mappingOutput, seqFromPdbMapping);
+
return siftsMapping;
}
public HashMap<Integer, int[]> getGreedyMapping(String entityId,
SequenceI seq, java.io.PrintStream os) throws SiftsException
{
- List<Integer> omitNonObserved = new ArrayList<Integer>();
- int nonObservedShiftIndex = 0;
+ List<Integer> omitNonObserved = new ArrayList<>();
+ int nonObservedShiftIndex = 0, pdbeNonObserved = 0;
// System.out.println("Generating mappings for : " + entityId);
Entity entity = null;
entity = getEntityById(entityId);
TreeMap<Integer, String> resNumMap = new TreeMap<Integer, String>();
List<Segment> segments = entity.getSegment();
SegmentHelperPojo shp = new SegmentHelperPojo(seq, mapping, resNumMap,
- omitNonObserved, nonObservedShiftIndex);
+ omitNonObserved, nonObservedShiftIndex, pdbeNonObserved);
processSegments(segments, shp);
try
{
{
throw new SiftsException("SIFTS mapping failed");
}
+ // also construct a mapping object between the seq-coord sys and the PDB
+ // seq's coord sys
Integer[] keys = mapping.keySet().toArray(new Integer[0]);
Arrays.sort(keys);
seqStart = keys[0];
seqEnd = keys[keys.length - 1];
-
+ List<int[]> from = new ArrayList<>(), to = new ArrayList<>();
+ int[] _cfrom = null, _cto = null;
String matchedSeq = originalSeq;
- if (seqStart != UNASSIGNED)
+ if (seqStart != UNASSIGNED) // fixme! seqStart can map to -1 for a pdb
+ // sequence that starts <-1
{
+ for (int seqps : keys)
+ {
+ int pdbpos = mapping.get(seqps)[PDBE_POS];
+ if (pdbpos == UNASSIGNED)
+ {
+ // not correct - pdbpos might be -1, but leave it for now
+ continue;
+ }
+ if (_cfrom == null || seqps != _cfrom[1] + 1)
+ {
+ _cfrom = new int[] { seqps, seqps };
+ from.add(_cfrom);
+ _cto = null; // discontinuity
+ }
+ else
+ {
+ _cfrom[1] = seqps;
+ }
+ if (_cto == null || pdbpos != 1 + _cto[1])
+ {
+ _cto = new int[] { pdbpos, pdbpos };
+ to.add(_cto);
+ }
+ else
+ {
+ _cto[1] = pdbpos;
+ }
+ }
+ _cfrom = new int[from.size() * 2];
+ _cto = new int[to.size() * 2];
+ int p = 0;
+ for (int[] range : from)
+ {
+ _cfrom[p++] = range[0];
+ _cfrom[p++] = range[1];
+ }
+ ;
+ p = 0;
+ for (int[] range : to)
+ {
+ _cto[p++] = range[0];
+ _cto[p++] = range[1];
+ }
+ ;
+
+ seqFromPdbMapping = new jalview.datamodel.Mapping(null, _cto, _cfrom,
+ 1, 1);
pdbStart = mapping.get(seqStart)[PDB_RES_POS];
pdbEnd = mapping.get(seqEnd)[PDB_RES_POS];
int orignalSeqStart = seq.getStart();
if (orignalSeqStart >= 1)
{
- int subSeqStart = (seqStart >= orignalSeqStart) ? seqStart
- - orignalSeqStart : 0;
+ int subSeqStart = (seqStart >= orignalSeqStart)
+ ? seqStart - orignalSeqStart
+ : 0;
int subSeqEnd = seqEnd - (orignalSeqStart - 1);
subSeqEnd = originalSeq.length() < subSeqEnd ? originalSeq.length()
: subSeqEnd;
TreeMap<Integer, String> resNumMap = shp.getResNumMap();
List<Integer> omitNonObserved = shp.getOmitNonObserved();
int nonObservedShiftIndex = shp.getNonObservedShiftIndex();
+ int pdbeNonObservedCount = shp.getPdbeNonObserved();
+ int firstPDBResNum = UNASSIGNED;
for (Segment segment : segments)
{
// System.out.println("Mapping segments : " + segment.getSegId() + "\\"s
List<Residue> residues = segment.getListResidue().getResidue();
for (Residue residue : residues)
{
+ boolean isObserved = isResidueObserved(residue);
+ int pdbeIndex = getLeadingIntegerValue(residue.getDbResNum(),
+ UNASSIGNED);
int currSeqIndex = UNASSIGNED;
List<CrossRefDb> cRefDbs = residue.getCrossRefDb();
CrossRefDb pdbRefDb = null;
if (cRefDb.getDbSource().equalsIgnoreCase(DBRefSource.PDB))
{
pdbRefDb = cRefDb;
- }
- if (cRefDb.getDbCoordSys()
- .equalsIgnoreCase(seqCoordSys.getName())
- && isAccessionMatched(cRefDb.getDbAccessionId()))
- {
- String resNumIndexString = cRefDb.getDbResNum()
- .equalsIgnoreCase("None") ? String.valueOf(UNASSIGNED)
- : cRefDb.getDbResNum();
- try
+ if (firstPDBResNum == UNASSIGNED)
{
- currSeqIndex = Integer.valueOf(resNumIndexString);
- } catch (NumberFormatException nfe)
+ firstPDBResNum = getLeadingIntegerValue(cRefDb.getDbResNum(),
+ UNASSIGNED);
+ }
+ else
{
- currSeqIndex = Integer.valueOf(resNumIndexString
- .split("[a-zA-Z]")[0]);
- continue;
+ if (isObserved)
+ {
+ // after we find the first observed residue we just increment
+ firstPDBResNum++;
+ }
}
+ }
+ if (cRefDb.getDbCoordSys().equalsIgnoreCase(seqCoordSys.getName())
+ && isAccessionMatched(cRefDb.getDbAccessionId()))
+ {
+ currSeqIndex = getLeadingIntegerValue(cRefDb.getDbResNum(),
+ UNASSIGNED);
if (pdbRefDb != null)
{
break;// exit loop if pdb and uniprot are already found
}
}
}
- if (currSeqIndex == UNASSIGNED)
+ if (!isObserved)
{
- continue;
+ ++pdbeNonObservedCount;
}
- if (currSeqIndex >= seq.getStart() && currSeqIndex <= seq.getEnd())
+ if (seqCoordSys == seqCoordSys.PDB) // FIXME: is seqCoordSys ever PDBe
+ // ???
{
- int resNum;
- try
- {
- resNum = (pdbRefDb == null) ? Integer.valueOf(residue
- .getDbResNum()) : Integer.valueOf(pdbRefDb
- .getDbResNum());
- } catch (NumberFormatException nfe)
+ // if the sequence has a primary reference to the PDB, then we are
+ // dealing with a sequence extracted directly from the PDB. In that
+ // case, numbering is PDBe - non-observed residues
+ currSeqIndex = seq.getStart() - 1 + pdbeIndex;
+ }
+ if (!isObserved)
+ {
+ if (seqCoordSys != CoordinateSys.UNIPROT) // FIXME: PDB or PDBe only
+ // here
{
- resNum = (pdbRefDb == null) ? Integer.valueOf(residue
- .getDbResNum()) : Integer.valueOf(pdbRefDb
- .getDbResNum().split("[a-zA-Z]")[0]);
- continue;
+ // mapping to PDB or PDBe so we need to bookkeep for the
+ // non-observed
+ // SEQRES positions
+ omitNonObserved.add(currSeqIndex);
+ ++nonObservedShiftIndex;
}
+ }
+ if (currSeqIndex == UNASSIGNED)
+ {
+ // change in logic - unobserved residues with no currSeqIndex
+ // corresponding are still counted in both nonObservedShiftIndex and
+ // pdbeIndex...
+ continue;
+ }
+ // if (currSeqIndex >= seq.getStart() && currSeqIndex <= seqlength) //
+ // true
+ // numbering
+ // is
+ // not
+ // up
+ // to
+ // seq.getEnd()
+ {
+
+ int resNum = (pdbRefDb == null)
+ ? getLeadingIntegerValue(residue.getDbResNum(),
+ UNASSIGNED)
+ : getLeadingIntegerValue(pdbRefDb.getDbResNum(),
+ UNASSIGNED);
- if (isResidueObserved(residue)
- || seqCoordSys == CoordinateSys.UNIPROT)
+ if (isObserved)
{
char resCharCode = ResidueProperties
.getSingleCharacterCode(ResidueProperties
.getCanonicalAminoAcid(residue.getDbResName()));
resNumMap.put(currSeqIndex, String.valueOf(resCharCode));
+
+ int[] mappingcols = new int[] { Integer.valueOf(resNum),
+ UNASSIGNED, isObserved ? firstPDBResNum : UNASSIGNED };
+
+ mapping.put(currSeqIndex - nonObservedShiftIndex, mappingcols);
}
- else
- {
- omitNonObserved.add(currSeqIndex);
- ++nonObservedShiftIndex;
- }
- mapping.put(currSeqIndex - nonObservedShiftIndex, new int[] {
- Integer.valueOf(resNum), UNASSIGNED });
}
}
}
}
/**
+ * Get the leading integer part of a string that begins with an integer.
+ *
+ * @param input
+ * - the string input to process
+ * @param failValue
+ * - value returned if unsuccessful
+ * @return
+ */
+ static int getLeadingIntegerValue(String input, int failValue)
+ {
+ if (input == null)
+ {
+ return failValue;
+ }
+ String[] parts = input.split("(?=\\D)(?<=\\d)");
+ if (parts != null && parts.length > 0 && parts[0].matches("[0-9]+"))
+ {
+ return Integer.valueOf(parts[0]);
+ }
+ return failValue;
+ }
+
+ /**
*
* @param chainId
* Target chain to populate mapping of its atom positions.
*/
public Entity getEntityByMostOptimalMatchedId(String chainId)
{
- // System.out.println("---> advanced greedy entityId matching block entered..");
+ // System.out.println("---> advanced greedy entityId matching block
+ // entered..");
List<Entity> entities = siftsEntry.getEntity();
SiftsEntitySortPojo[] sPojo = new SiftsEntitySortPojo[entities.size()];
int count = 0;
return null;
}
- private class SiftsEntitySortPojo implements
- Comparable<SiftsEntitySortPojo>
+ private class SiftsEntitySortPojo
+ implements Comparable<SiftsEntitySortPojo>
{
public String entityId;
private int nonObservedShiftIndex;
- public SegmentHelperPojo(SequenceI seq,
- HashMap<Integer, int[]> mapping,
+ /**
+ * count of number of 'not observed' positions in the PDB record's SEQRES
+ * (total number of residues with coordinates == length(SEQRES) -
+ * pdbeNonObserved
+ */
+ private int pdbeNonObserved;
+
+ public SegmentHelperPojo(SequenceI seq, HashMap<Integer, int[]> mapping,
TreeMap<Integer, String> resNumMap,
- List<Integer> omitNonObserved, int nonObservedShiftIndex)
+ List<Integer> omitNonObserved, int nonObservedShiftIndex,
+ int pdbeNonObserved)
{
setSeq(seq);
setMapping(mapping);
setResNumMap(resNumMap);
setOmitNonObserved(omitNonObserved);
setNonObservedShiftIndex(nonObservedShiftIndex);
+ setPdbeNonObserved(pdbeNonObserved);
+
+ }
+
+ public void setPdbeNonObserved(int pdbeNonObserved2)
+ {
+ this.pdbeNonObserved = pdbeNonObserved2;
+ }
+
+ public int getPdbeNonObserved()
+ {
+ return pdbeNonObserved;
}
public SequenceI getSeq()
{
this.nonObservedShiftIndex = nonObservedShiftIndex;
}
+
}
@Override
- public StringBuffer getMappingOutput(MappingOutputPojo mp)
+ public StringBuilder getMappingOutput(MappingOutputPojo mp)
throws SiftsException
{
String seqRes = mp.getSeqResidue();
int nochunks = ((seqRes.length()) / len)
+ ((seqRes.length()) % len > 0 ? 1 : 0);
// output mappings
- StringBuffer output = new StringBuffer();
+ StringBuilder output = new StringBuilder(512);
output.append(NEWLINE);
- output.append("Sequence \u27f7 Structure mapping details").append(
- NEWLINE);
+ output.append("Sequence \u27f7 Structure mapping details")
+ .append(NEWLINE);
output.append("Method: SIFTS");
output.append(NEWLINE).append(NEWLINE);
output.append(String.valueOf(pdbEnd));
output.append(NEWLINE).append(NEWLINE);
+ ScoreMatrix pam250 = ScoreModels.getInstance().getPam250();
int matchedSeqCount = 0;
for (int j = 0; j < nochunks; j++)
{
// Print the first aligned sequence
- output.append(new Format("%" + (maxid) + "s").form(seqName)).append(
- " ");
+ output.append(new Format("%" + (maxid) + "s").form(seqName))
+ .append(" ");
for (int i = 0; i < len; i++)
{
output.append(NEWLINE);
output.append(new Format("%" + (maxid) + "s").form(" ")).append(" ");
- // Print out the matching chars
+ /*
+ * Print out the match symbols:
+ * | for exact match (ignoring case)
+ * . if PAM250 score is positive
+ * else a space
+ */
for (int i = 0; i < len; i++)
{
try
{
if ((i + (j * len)) < seqRes.length())
{
- boolean sameChar = Comparison.isSameResidue(
- seqRes.charAt(i + (j * len)),
- strRes.charAt(i + (j * len)), false);
- if (sameChar
- && !jalview.util.Comparison.isGap(seqRes.charAt(i
- + (j * len))))
+ char c1 = seqRes.charAt(i + (j * len));
+ char c2 = strRes.charAt(i + (j * len));
+ boolean sameChar = Comparison.isSameResidue(c1, c2, false);
+ if (sameChar && !Comparison.isGap(c1))
{
matchedSeqCount++;
output.append("|");
}
else if (type.equals("pep"))
{
- if (ResidueProperties.getPAM250(seqRes.charAt(i + (j * len)),
- strRes.charAt(i + (j * len))) > 0)
+ if (pam250.getPairwiseScore(c1, c2) > 0)
{
output.append(".");
}
{
throw new SiftsException(">>> Low PID detected for SIFTs mapping...");
}
- output.append("Length of alignment = " + seqRes.length()).append(
- NEWLINE);
+ output.append("Length of alignment = " + seqRes.length())
+ .append(NEWLINE);
output.append(new Format("Percentage ID = %2.2f").form(pid));
return output;
}
git://source.jalview.org / jalview.git / blobdiff