use of ubic.gemma.persistence.service.association.coexpression.LinkCreator in project Gemma by PavlidisLab.
the class Gene2GeneCoexpressionServiceTest method setup.
@Before
public void setup() {
Taxon mouseTaxon = taxonS.findByCommonName("mouse");
firstGene = Gene.Factory.newInstance();
firstGene.setName("test_gene2geneCoexpression");
firstGene.setTaxon(mouseTaxon);
firstGene = geneS.create(firstGene);
Gene secondGene = Gene.Factory.newInstance();
secondGene.setName("test_gene2geneCoexpression2");
secondGene.setTaxon(mouseTaxon);
secondGene = geneS.create(secondGene);
List<NonPersistentNonOrderedCoexpLink> links = new ArrayList<>();
links.add(new NonPersistentNonOrderedCoexpLink(MouseGeneCoExpression.Factory.newInstance(0.9, secondGene.getId(), firstGene.getId())));
ee = this.getTestPersistentBasicExpressionExperiment();
Set<Gene> genesTested = new HashSet<>();
genesTested.add(firstGene);
genesTested.add(secondGene);
g2gCoexpressionService.createOrUpdate(ee, links, new LinkCreator(mouseTaxon), genesTested);
}
use of ubic.gemma.persistence.service.association.coexpression.LinkCreator in project Gemma by PavlidisLab.
the class LinkAnalysisPersisterImpl method getLinkCreator.
private LinkCreator getLinkCreator(LinkAnalysis la) {
Taxon taxon = la.getTaxon();
LinkCreator c;
c = new LinkCreator(taxon);
return c;
}
use of ubic.gemma.persistence.service.association.coexpression.LinkCreator in project Gemma by PavlidisLab.
the class LinkAnalysisPersisterImpl method saveLinks.
/**
* @return how many links were saved
*/
private int saveLinks(LinkAnalysis la, ObjectArrayList links) {
LinkCreator c = this.getLinkCreator(la);
int selfLinksSkipped = 0;
int duplicateLinksSkipped = 0;
Set<Gene> genesWithLinks = new HashSet<>();
Set<NonPersistentNonOrderedCoexpLink> linksForDb = new HashSet<>();
for (int i = 0, n = links.size(); i < n; i++) {
Object val = links.getQuick(i);
if (val == null)
continue;
Link m = (Link) val;
Double w = m.getWeight();
int x = m.getx();
int y = m.gety();
CompositeSequence p1 = la.getProbe(x);
CompositeSequence p2 = la.getProbe(y);
/*
* we have to deal with all the possible genes pairs, if probes map to more than one pair. A single pair of
* probes could result in more than one link. This assumes that preprocessing of the data allowed retention
* of probes that map to more than one gene.
*/
for (Gene g1 : la.getProbeToGeneMap().get(p1)) {
boolean g1HasLinks = false;
for (Gene g2 : la.getProbeToGeneMap().get(p2)) {
if (g1.equals(g2)) {
selfLinksSkipped++;
continue;
}
NonPersistentNonOrderedCoexpLink link = new NonPersistentNonOrderedCoexpLink(this.initCoexp(w, c, g1, g2));
if (linksForDb.contains(link)) {
/*
* This happens if there is more than one probe retained for a gene (or both genes) and the
* coexpression shows up more than once (different pair of probes, same genes).
*/
if (LinkAnalysisPersisterImpl.log.isDebugEnabled())
LinkAnalysisPersisterImpl.log.debug("Skipping duplicate: " + link);
duplicateLinksSkipped++;
continue;
/*
* FIXME what do we do when a pair of genes is both positively and negatively correlated in the
* same experiment? Currently they are both kept, but if we go to a completely gene-based
* analysis we wouldn't do that, so it's an inconsistency;
*/
}
if (LinkAnalysisPersisterImpl.log.isDebugEnabled()) {
LinkAnalysisPersisterImpl.log.debug("Adding : " + link);
}
linksForDb.add(link);
g1HasLinks = true;
genesWithLinks.add(g2);
}
if (g1HasLinks)
genesWithLinks.add(g1);
}
if (i > 0 && i % 200000 == 0) {
LinkAnalysisPersisterImpl.log.info(i + " links checked");
}
}
if (selfLinksSkipped > 0) {
LinkAnalysisPersisterImpl.log.info(selfLinksSkipped + " self-links skipped");
}
if (duplicateLinksSkipped > 0) {
LinkAnalysisPersisterImpl.log.info(duplicateLinksSkipped + " duplicate links skipped (likely cause: more than one probe supporting the same link)");
}
if (linksForDb.isEmpty()) {
throw new RuntimeException("No links left!");
}
LinkAnalysisPersisterImpl.log.info(linksForDb.size() + " links ready for saving to db");
if (!la.getGenesTested().containsAll(genesWithLinks))
throw new AssertionError();
/*
* Do the actual database writing. It's a good idea to do this part in one (big) transaction. Note that even if
* there are no links, we still update the "genes tested" information.
*/
this.gene2GeneCoexpressionService.createOrUpdate(la.getExpressionExperiment(), new ArrayList<>(linksForDb), c, la.getGenesTested());
/*
* Update the meta-data about the analysis
*/
CoexpressionAnalysis analysisObj = la.getAnalysisObj();
assert analysisObj.getId() != null;
analysisObj.setNumberOfElementsAnalyzed(la.getGenesTested().size());
analysisObj.setNumberOfLinks(linksForDb.size());
coexpressionAnalysisService.update(analysisObj);
return linksForDb.size();
/*
* Updating node degree cannot be done here, since we need to know the support. We have to do that
* "periodically" if we want it available in summary form.
*/
}
Aggregations