*/
package jalview.ext.ensembl;
+import java.io.IOException;
+import java.net.MalformedURLException;
+import java.net.URL;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.List;
+import java.util.Map;
+
+import org.json.simple.parser.ParseException;
+
import jalview.analysis.AlignmentUtils;
import jalview.analysis.Dna;
-import jalview.bin.Cache;
+import jalview.bin.Console;
import jalview.datamodel.Alignment;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.DBRefEntry;
import jalview.util.DBRefUtils;
import jalview.util.IntRangeComparator;
import jalview.util.MapList;
-import jalview.util.Platform;
-
-import java.io.IOException;
-import java.net.MalformedURLException;
-import java.net.URL;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.Collections;
-import java.util.List;
-import java.util.Map;
-
-import org.json.simple.parser.ParseException;
/**
* Base class for Ensembl sequence fetchers
String msg = "Aborting ID retrieval after " + v
+ " chunks. Unexpected problem (" + r.getLocalizedMessage()
+ ")";
- System.err.println(msg);
+ jalview.bin.Console.errPrintln(msg);
r.printStackTrace();
break;
}
* fetch and transfer genomic sequence features,
* fetch protein product and add as cross-reference
*/
- for (String accId : allIds)
+ for (int i = 0, n = allIds.size(); i < n; i++)
{
- addFeaturesAndProduct(accId, alignment);
+ addFeaturesAndProduct(allIds.get(i), alignment);
}
- for (SequenceI seq : alignment.getSequences())
+ List<SequenceI> seqs = alignment.getSequences();
+ for (int i = 0, n = seqs.size(); i < n; i++)
{
- getCrossReferences(seq);
+ getCrossReferences(seqs.get(i));
}
return alignment;
*/
SequenceI genomicSequence = null;
EnsemblFeatures gffFetcher = new EnsemblFeatures(getDomain());
- EnsemblFeatureType[] features = getFeaturesToFetch();
+ EnsemblFeatureType[] features = getFeaturesToFetch();
+
+ // Platform.timeCheck("ESP.getsequencerec1", Platform.TIME_MARK);
+
AlignmentI geneFeatures = gffFetcher.getSequenceRecords(accId,
features);
if (geneFeatures != null && geneFeatures.getHeight() > 0)
{
genomicSequence = geneFeatures.getSequenceAt(0);
}
+
+ // Platform.timeCheck("ESP.getsequencerec2", Platform.TIME_MARK);
+
if (genomicSequence != null)
{
/*
* fetch and map protein product, and add it as a cross-reference
* of the retrieved sequence
*/
+ // Platform.timeCheck("ESP.transferFeatures", Platform.TIME_MARK);
addProteinProduct(querySeq);
}
}
} catch (IOException e)
{
- System.err.println(
+ jalview.bin.Console.errPrintln(
"Error transferring Ensembl features: " + e.getMessage());
}
+ // Platform.timeCheck("ESP.addfeat done", Platform.TIME_MARK);
}
/**
String accId = querySeq.getName();
try
{
- System.out.println("Adding protein product for " + accId);
+ jalview.bin.Console.outPrintln("Adding protein product for " + accId);
AlignmentI protein = new EnsemblProtein(getDomain())
.getSequenceRecords(accId);
if (protein == null || protein.getHeight() == 0)
{
- System.out.println("No protein product found for " + accId);
+ jalview.bin.Console.outPrintln("No protein product found for " + accId);
return;
}
SequenceI proteinSeq = protein.getSequenceAt(0);
List<DBRefEntry> uprots = DBRefUtils.selectRefs(ds.getDBRefs(),
new String[]
{ DBRefSource.UNIPROT });
- List<DBRefEntry> upxrefs = DBRefUtils.selectRefs(querySeq.getDBRefs(),
- new String[]
+ List<DBRefEntry> upxrefs = DBRefUtils
+ .selectRefs(querySeq.getDBRefs(), new String[]
{ DBRefSource.UNIPROT });
if (uprots != null)
{
if (upx.size() > 1)
{
- Cache.log.warn(
+ Console.warn(
"Implementation issue - multiple uniprot acc on product sequence.");
}
}
* copy exon features to protein, compute peptide variants from dna
* variants and add as features on the protein sequence ta-da
*/
- AlignmentUtils.computeProteinFeatures(querySeq, proteinSeq,
- mapList);
+ // JAL-3187 render on the fly instead
+ // AlignmentUtils.computeProteinFeatures(querySeq, proteinSeq, mapList);
}
} catch (Exception e)
{
*/
protected void getCrossReferences(SequenceI seq)
{
+
+ // Platform.timeCheck("ESP. getdataseq ", Platform.TIME_MARK);
+
while (seq.getDatasetSequence() != null)
{
seq = seq.getDatasetSequence();
}
+
+ // Platform.timeCheck("ESP. getxref ", Platform.TIME_MARK);
+
EnsemblXref xrefFetcher = new EnsemblXref(getDomain(), getDbSource(),
getEnsemblDataVersion());
List<DBRefEntry> xrefs = xrefFetcher.getCrossReferences(seq.getName());
-
+
for (int i = 0, n = xrefs.size(); i < n; i++)
{
- // BH 2019.01.25 this next method was taking 174 ms PER addition for a 266-reference example.
- // DBRefUtils.ensurePrimaries(seq)
- // was at the end of seq.addDBRef, so executed after ever addition!
- // This method was moved to seq.getPrimaryDBRefs()
+ // Platform.timeCheck("ESP. getxref + " + (i) + "/" + n,
+ // Platform.TIME_MARK);
+ // BH 2019.01.25 this next method was taking 174 ms PER addition for a
+ // 266-reference example.
+ // DBRefUtils.ensurePrimaries(seq)
+ // was at the end of seq.addDBRef, so executed after ever addition!
+ // This method was moved to seq.getPrimaryDBRefs()
seq.addDBRef(xrefs.get(i));
}
+
+ // jalview.bin.Console.outPrintln("primaries are " + seq.getPrimaryDBRefs().toString());
/*
* and add a reference to itself
*/
+
+ // Platform.timeCheck("ESP. getxref self ", Platform.TIME_MARK);
+
DBRefEntry self = new DBRefEntry(getDbSource(), getEnsemblDataVersion(),
- seq.getName());
+ seq.getName());
+
+ // Platform.timeCheck("ESP. getxref self add ", Platform.TIME_MARK);
+
seq.addDBRef(self);
+
+ // Platform.timeCheck("ESP. seqprox done ", Platform.TIME_MARK);
}
/**
inProgress = false;
throw new JalviewException("ENSEMBL Rest API not available.");
}
+ // Platform.timeCheck("EnsemblSeqProx.fetchSeq ", Platform.TIME_MARK);
+
List<SequenceI> seqs = parseSequenceJson(ids);
if (seqs == null)
- return alignment;
+ return alignment;
if (seqs.isEmpty())
{
if (seqs.size() != ids.size())
{
- System.out.println(String.format(
+ jalview.bin.Console.outPrintln(String.format(
"Only retrieved %d sequences for %d query strings",
seqs.size(), ids.size()));
}
* for now, assumes only one sequence returned; refactor if needed
* in future to handle a JSONArray with more than one
*/
- Map<String, Object> val = (Map<String, Object>) getJSON(null, ids, -1, MODE_MAP, null);
+ // Platform.timeCheck("ENS seqproxy", Platform.TIME_MARK);
+ Map<String, Object> val = (Map<String, Object>) getJSON(null, ids, -1,
+ MODE_MAP, null);
if (val == null)
- return null;
+ return null;
Object s = val.get("desc");
String desc = s == null ? null : s.toString();
s = val.get("id");
result.add(sequence);
} catch (ParseException | IOException e)
{
- System.err.println("Error processing JSON response: " + e.toString());
+ jalview.bin.Console.errPrintln("Error processing JSON response: " + e.toString());
// ignore
}
+ // Platform.timeCheck("ENS seqproxy2", Platform.TIME_MARK);
return result;
}
if (regions.isEmpty())
{
- System.out.println("Failed to identify target sequence for " + accId
+ jalview.bin.Console.outPrintln("Failed to identify target sequence for " + accId
+ " from genomic features");
return null;
}
protected abstract List<SequenceFeature> getIdentifyingFeatures(
SequenceI seq, String accId);
+ int bhtest = 0;
+
/**
* Transfers the sequence feature to the target sequence, locating its start
* and end range based on the mapping. Features which do not overlap the
if (mappedRange != null)
{
+ // Platform.timeCheck(null, Platform.TIME_SET);
String group = sf.getFeatureGroup();
if (".".equals(group))
{
}
int newBegin = Math.min(mappedRange[0], mappedRange[1]);
int newEnd = Math.max(mappedRange[0], mappedRange[1]);
+ // Platform.timeCheck(null, Platform.TIME_MARK);
+ bhtest++;
+ // 280 ms/1000 here:
SequenceFeature copy = new SequenceFeature(sf, newBegin, newEnd,
group, sf.getScore());
+ // 0.175 ms here:
targetSequence.addSequenceFeature(copy);
/*
String comp = complement.toString();
sf.setValue(Gff3Helper.ALLELES, comp);
sf.setDescription(comp);
-
- /*
- * replace value of "alleles=" in sf.ATTRIBUTES as well
- * so 'output as GFF' shows reverse complement alleles
- */
- String atts = sf.getAttributes();
- if (atts != null)
- {
- atts = atts.replace(Gff3Helper.ALLELES + "=" + alleles,
- Gff3Helper.ALLELES + "=" + comp);
- sf.setAttributes(atts);
- }
}
/**
return false;
}
-// long start = System.currentTimeMillis();
+ // long start = System.currentTimeMillis();
List<SequenceFeature> sfs = sourceSequence.getFeatures()
.getPositionalFeatures();
MapList mapping = getGenomicRangesFromFeatures(sourceSequence,
return false;
}
+ // Platform.timeCheck("ESP. xfer " + sfs.size(), Platform.TIME_MARK);
+
boolean result = transferFeatures(sfs, targetSequence, mapping,
accessionId);
-// System.out.println("transferFeatures (" + (sfs.size()) + " --> "
-// + targetSequence.getFeatures().getFeatureCount(true) + ") to "
-// + targetSequence.getName() + " took "
-// + (System.currentTimeMillis() - start) + "ms");
+ // jalview.bin.Console.outPrintln("transferFeatures (" + (sfs.size()) + " --> "
+ // + targetSequence.getFeatures().getFeatureCount(true) + ") to "
+ // + targetSequence.getName() + " took "
+ // + (System.currentTimeMillis() - start) + "ms");
return result;
}
SequenceFeatures.sortFeatures(sfs, forwardStrand);
boolean transferred = false;
- for (SequenceFeature sf : sfs)
+
+ for (int i = 0, n = sfs.size(); i < n; i++)
{
+
+ // if ((i%1000) == 0) {
+ //// Platform.timeCheck("Feature " + bhtest, Platform.TIME_GET);
+ // Platform.timeCheck("ESP. xferFeature + " + (i) + "/" + n,
+ // Platform.TIME_MARK);
+ // }
+
+ SequenceFeature sf = sfs.get(i);
if (retainFeature(sf, parentId))
{
transferFeature(sf, targetSequence, mapping, forwardStrand);
transferred = true;
}
}
+
return transferred;
}
protected boolean featureMayBelong(SequenceFeature sf, String identifier)
{
String parent = (String) sf.getValue(PARENT);
- if (parent != null
- && !parent.equalsIgnoreCase(identifier))
+ if (parent != null && !parent.equalsIgnoreCase(identifier))
{
// this genomic feature belongs to a different transcript
return false;
git://source.jalview.org / jalview.git / blobdiff