Menu
▾
▴
[Gmod-apollo-cmts] SF.net SVN: gmod:[23134] apollo/trunk/src/java/apollo/dataadapter
|
From: <gk...@us...> - 2010-05-05 14:17:05
|
Revision: 23134
http://gmod.svn.sourceforge.net/gmod/?rev=23134&view=rev
Author: gk_fan
Date: 2010-05-05 14:16:49 +0000 (Wed, 05 May 2010)
Log Message:
-----------
+ Fixed GFF3 input from URL when run from the command line
Modified Paths:
--------------
apollo/trunk/src/java/apollo/dataadapter/DataInput.java
apollo/trunk/src/java/apollo/dataadapter/gff3/GFF3Adapter.java
Modified: apollo/trunk/src/java/apollo/dataadapter/DataInput.java
===================================================================
--- apollo/trunk/src/java/apollo/dataadapter/DataInput.java 2010-05-05 13:26:27 UTC (rev 23133)
+++ apollo/trunk/src/java/apollo/dataadapter/DataInput.java 2010-05-05 14:16:49 UTC (rev 23134)
@@ -124,14 +124,27 @@
return getInputString() != null;
}
- public boolean isFile() { return getType() == DataInputType.FILE; }
-
+ public boolean isFile() {
+ return getType() == DataInputType.FILE;
+ }
+
public String getFilename() {
if (!isFile())
return null;
return getInputString();
}
+
+ public boolean isUrl() {
+ return getType() == DataInputType.URL;
+ }
+ public String getUrl() {
+ if (!isUrl()) {
+ return null;
+ }
+ return getInputString();
+ }
+
/** Returns true if DataInputType is BASEPAIR_RANGE */
public boolean isRegion() {
return getType() == DataInputType.BASEPAIR_RANGE;
Modified: apollo/trunk/src/java/apollo/dataadapter/gff3/GFF3Adapter.java
===================================================================
--- apollo/trunk/src/java/apollo/dataadapter/gff3/GFF3Adapter.java 2010-05-05 13:26:27 UTC (rev 23133)
+++ apollo/trunk/src/java/apollo/dataadapter/gff3/GFF3Adapter.java 2010-05-05 14:16:49 UTC (rev 23134)
@@ -122,7 +122,12 @@
*/
public void setDataInput(apollo.dataadapter.DataInput di)
{
- setGffFilename(di.getFilename());
+ if (di.isFile()) {
+ setGffFilename(di.getFilename());
+ }
+ else if (di.isUrl()) {
+ setGffFilename(di.getUrl());
+ }
if (di.hasSequenceFilename()) {
setSeqFilename(di.getSequenceFilename());
}
@@ -238,6 +243,7 @@
*/
private void writeSeqs(PrintWriter out, List<SequenceI> seqs)
{
+ out.println("##FASTA");
for (SequenceI seq : seqs) {
if (seq.getLength() > 0) {
out.print(FastaFile.print(seq));
@@ -597,169 +603,168 @@
Map<String, int []> cdsCoords, CurationSet curation)
throws ApolloAdapterException
{
- Map<String, FeatureSetI[]> resultSets =
- new TreeMap<String, FeatureSetI[]>();
- Set<String> errors = new HashSet<String>();
- Set<GFF3Data> matches = new HashSet<GFF3Data>();
- for (Map.Entry<String, GFF3Data> mapEntry : gffMap.entrySet()) {
- GFF3Data data = mapEntry.getValue();
- GFF3Entry gff = data.getEntry();
- SeqFeatureI feat = data.getFeat();
- feat.setRefSequence(curation.getRefSequence());
- if (!gff.isAnnotation()) {
- processEvidence(feat);
- }
- // add matches for post processing if they're missing match_part features
- if (gff.getType().equals("match")) {
- matches.add(data);
- }
- boolean added = false;
- List<String> pids = gff.getAttributeValues("Parent");
- if (pids != null) {
- int pnum = 0;
- for (String pid : pids) {
- GFF3Data pdata = gffMap.get(pid);
- if (gff.isAnnotation()) {
- SeqFeatureI parent = pdata.getFeat();
- if (!feat.isTranscript() && !feat.isExon()) {
- if (!errors.contains(feat.getName())) {
- IOUtil.errorDialog("Only transcripts and exons can be children of annotations: " +
- feat.getName() + "[" + feat.getFeatureType() + "]");
- errors.add(feat.getName());
- }
- added = true;
- continue;
+ Map<String, FeatureSetI[]> resultSets = new TreeMap<String, FeatureSetI[]>();
+ Set<String> errors = new HashSet<String>();
+ Set<GFF3Data> matches = new HashSet<GFF3Data>();
+ for (Map.Entry<String, GFF3Data> mapEntry : gffMap.entrySet()) {
+ GFF3Data data = mapEntry.getValue();
+ GFF3Entry gff = data.getEntry();
+ SeqFeatureI feat = data.getFeat();
+ feat.setRefSequence(curation.getRefSequence());
+ if (!gff.isAnnotation()) {
+ processEvidence(feat);
+ }
+ // add matches for post processing if they're missing match_part features
+ if (gff.getType().equals("match")) {
+ matches.add(data);
+ }
+ boolean added = false;
+ List<String> pids = gff.getAttributeValues("Parent");
+ if (pids != null) {
+ int pnum = 0;
+ for (String pid : pids) {
+ GFF3Data pdata = gffMap.get(pid);
+ if (gff.isAnnotation()) {
+ SeqFeatureI parent = pdata.getFeat();
+ if (!feat.isTranscript() && !feat.isExon()) {
+ if (!errors.contains(feat.getName())) {
+ IOUtil.errorDialog("Only transcripts and exons can be children of annotations: " +
+ feat.getName() + "[" + feat.getFeatureType() + "]");
+ errors.add(feat.getName());
}
- if (feat.isExon() && !parent.isTranscript()) {
- if (!errors.contains(feat.getName())) {
- IOUtil.errorDialog("Exons can only be children of transcripts: " + feat.getName() + " -> " +
- parent.getName() + "[" + parent.getFeatureType() + "]");
- errors.add(feat.getName());
- }
- added = true;
- continue;
- }
- }
- ++pnum;
- if (pdata == null) { //orphan
- logger.warn("Missing parent " + pid + " for " +
- gff.getName());
+ added = true;
continue;
}
- if (pnum > 1) {
- if (pnum == 2) {
- feat.setId(feat.getId() + "-1");
+ if (feat.isExon() && !parent.isTranscript()) {
+ if (!errors.contains(feat.getName())) {
+ IOUtil.errorDialog("Exons can only be children of transcripts: " + feat.getName() + " -> " +
+ parent.getName() + "[" + parent.getFeatureType() + "]");
+ errors.add(feat.getName());
}
- feat = feat.cloneFeature();
- feat.setId(feat.getId().replaceFirst("-\\d+$",
- "-" + pnum));
+ added = true;
+ continue;
}
- if (gff.isAnnotation()) {
- processAnnotation(data, pdata.getFeat(), cdsCoords);
+ }
+ ++pnum;
+ if (pdata == null) { //orphan
+ logger.warn("Missing parent " + pid + " for " +
+ gff.getName());
+ continue;
+ }
+ if (pnum > 1) {
+ if (pnum == 2) {
+ feat.setId(feat.getId() + "-1");
}
- try {
- pdata.getFeat().addFeature(feat);
- added = true;
- }
- catch (ClassCastException e) {
- /*
+ feat = feat.cloneFeature();
+ feat.setId(feat.getId().replaceFirst("-\\d+$",
+ "-" + pnum));
+ }
+ if (gff.isAnnotation()) {
+ processAnnotation(data, pdata.getFeat(), cdsCoords);
+ }
+ try {
+ pdata.getFeat().addFeature(feat);
+ added = true;
+ }
+ catch (ClassCastException e) {
+ /*
throw new ApolloAdapterException("Cannot add " + feat +
"(" + data.getEntry().getType() + ")" + " to " +
pdata.getFeat() + "(" + pdata.getEntry().getType() +
")");
- */
- logger.warn("Cannot add " + feat +
- "(" + data.getEntry().getType() + ")" + " to " +
- pdata.getFeat() + "(" + pdata.getEntry().getType() +
- ")");
- }
+ */
+ logger.warn("Cannot add " + feat +
+ "(" + data.getEntry().getType() + ")" + " to " +
+ pdata.getFeat() + "(" + pdata.getEntry().getType() +
+ ")");
}
}
- //a transcript can potentially not have a gene associated with it, in which
- //case we need to generate one (the GFF3 spec doesn't REQUIRE a transcript
- //to have a gene, which seems kind of silly to me...)
- else {
- if (!feat.isAnnotTop()) {
- if (gff.isAnnotation()) {
- logger.warn("No parent feature found for " + feat +
- "(" + gff.getType() + "). Generating dummy parent feature.");
- generateParent(feat, annots);
- added = true;
- }
+ }
+ //a transcript can potentially not have a gene associated with it, in which
+ //case we need to generate one (the GFF3 spec doesn't REQUIRE a transcript
+ //to have a gene, which seems kind of silly to me...)
+ else {
+ if (!feat.isAnnotTop()) {
+ if (gff.isAnnotation()) {
+ logger.warn("No parent feature found for " + feat +
+ "(" + gff.getType() + "). Generating dummy parent feature.");
+ generateParent(feat, annots);
+ added = true;
}
}
- if (!added) {
- if (gff.isAnnotation()) {
- // need to wrap one-level annotations in a AnnotationFeatureI otherwise Apollo isn't happy
- if (feat.isAnnotTop() && gff.isOneLevelAnnotation()) {
- AnnotatedFeatureI fs = new AnnotatedFeature();
- fs.addFeature(feat);
- fs.setStrand(feat.getStrand());
- fs.setId(feat.getId() + "-fs");
- fs.setFeatureType(feat.getFeatureType());
- feat = fs;
- }
- // arbitrarily add strand-less features to plus strand
- if (feat.getStrand() == 0) {
- annots.getForwardSet().addFeature(feat);
- }
- else {
- annots.addFeature(feat);
- }
+ }
+ if (!added) {
+ if (gff.isAnnotation()) {
+ // need to wrap one-level annotations in a AnnotationFeatureI otherwise Apollo isn't happy
+// if (feat.isAnnotTop() && gff.isOneLevelAnnotation()) {
+// AnnotatedFeatureI fs = new AnnotatedFeature();
+// fs.addFeature(feat);
+// fs.setStrand(feat.getStrand());
+// fs.setId(feat.getId() + "-fs");
+// fs.setFeatureType(feat.getFeatureType());
+// feat = fs;
+// }
+ // arbitrarily add strand-less features to plus strand
+ if (feat.getStrand() == 0) {
+ annots.getForwardSet().addFeature(feat);
}
else {
- processEvidence(feat);
- FeatureSetI []fsArray = resultSets.get(gff.getSource());
- if (fsArray == null) {
- fsArray = new FeatureSet[2];
- resultSets.put(gff.getSource(), fsArray);
- }
- int arrayIdx = gff.getStrand() == 1 ? 1 : 0;
- FeatureSetI fs = fsArray[arrayIdx];
- if (fs == null) {
- fs = new FeatureSet(gff.getSource(), gff.getStrand());
- fsArray[arrayIdx] = fs;
- }
- fs.addFeature(feat);
- /*
- results.addFeature(feat);
- */
+ annots.addFeature(feat);
}
}
- }
-
- // add FeaturePair objects to any matches with no match_part children
- for (GFF3Data data : matches) {
- SeqFeatureI match = data.getFeat();
- if (match.getFeatures().size() == 0) {
- GFF3Entry matchPart = new GFF3Entry(data.getEntry().toString());
- matchPart.setType("match_part");
- try {
- SeqFeatureI matchPartFeat = matchPart.createFeature(oboParser);
- processEvidence(matchPartFeat);
- match.addFeature(matchPartFeat);
+ else {
+ processEvidence(feat);
+ FeatureSetI []fsArray = resultSets.get(gff.getSource());
+ if (fsArray == null) {
+ fsArray = new FeatureSet[2];
+ resultSets.put(gff.getSource(), fsArray);
}
- catch (ClassNotFoundException e) {
- throw new ApolloAdapterException("Cannot locate Apollo datamodel class: " + e.getMessage());
+ int arrayIdx = gff.getStrand() == 1 ? 1 : 0;
+ FeatureSetI fs = fsArray[arrayIdx];
+ if (fs == null) {
+ fs = new FeatureSet(gff.getSource(), gff.getStrand());
+ fsArray[arrayIdx] = fs;
}
- catch (IllegalAccessException e) {
- throw new ApolloAdapterException(e);
- }
- catch (InstantiationException e) {
- throw new ApolloAdapterException("Cannot instantiate class: " + e.getMessage());
- }
+ fs.addFeature(feat);
+ /*
+ results.addFeature(feat);
+ */
}
}
+ }
- for (Map.Entry<String, FeatureSetI[]> e : resultSets.entrySet()) {
- if (e.getValue()[0] != null) {
- results.addFeature(e.getValue()[0]);
+ // add FeaturePair objects to any matches with no match_part children
+ for (GFF3Data data : matches) {
+ SeqFeatureI match = data.getFeat();
+ if (match.getFeatures().size() == 0) {
+ GFF3Entry matchPart = new GFF3Entry(data.getEntry().toString());
+ matchPart.setType("match_part");
+ try {
+ SeqFeatureI matchPartFeat = matchPart.createFeature(oboParser);
+ processEvidence(matchPartFeat);
+ match.addFeature(matchPartFeat);
}
- if (e.getValue()[1] != null) {
- results.addFeature(e.getValue()[1]);
+ catch (ClassNotFoundException e) {
+ throw new ApolloAdapterException("Cannot locate Apollo datamodel class: " + e.getMessage());
}
+ catch (IllegalAccessException e) {
+ throw new ApolloAdapterException(e);
+ }
+ catch (InstantiationException e) {
+ throw new ApolloAdapterException("Cannot instantiate class: " + e.getMessage());
+ }
}
- setTranslationEnds(gffMap, cdsCoords);
+ }
+
+ for (Map.Entry<String, FeatureSetI[]> e : resultSets.entrySet()) {
+ if (e.getValue()[0] != null) {
+ results.addFeature(e.getValue()[0]);
+ }
+ if (e.getValue()[1] != null) {
+ results.addFeature(e.getValue()[1]);
+ }
+ }
+ setTranslationEnds(gffMap, cdsCoords);
}
private void generateParent(SeqFeatureI feat, StrandedFeatureSetI annots)
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|