Search in sources :

Example 1 with ExpressionExperimentBibRefFinder

use of ubic.gemma.core.loader.entrez.pubmed.ExpressionExperimentBibRefFinder in project Gemma by PavlidisLab.

the class ExpressionExperimentPrimaryPubCli method doWork.

@Override
protected Exception doWork(String[] args) {
    Exception err = processCommandLine(args);
    if (err != null)
        return err;
    ExpressionExperimentService ees = this.getBean(ExpressionExperimentService.class);
    Persister ph = this.getPersisterHelper();
    PubMedXMLFetcher fetcher = new PubMedXMLFetcher();
    // collect some statistics
    Collection<String> nullPubCount = new ArrayList<>();
    Collection<String> samePubCount = new ArrayList<>();
    Collection<String> diffPubCount = new ArrayList<>();
    Collection<String> failedEe = new ArrayList<>();
    ExpressionExperimentBibRefFinder finder = new ExpressionExperimentBibRefFinder();
    for (BioAssaySet bioassay : expressionExperiments) {
        if (!(bioassay instanceof ExpressionExperiment)) {
            log.info(bioassay.getName() + " is not an ExpressionExperiment");
            continue;
        }
        ExpressionExperiment experiment = (ExpressionExperiment) bioassay;
        // if ( experiment.getPrimaryPublication() != null ) continue;
        if (experiment.getPrimaryPublication() == null) {
            log.warn(experiment + " has no existing primary publication");
        }
        experiment = ees.thawLite(experiment);
        // get from GEO or get from a file
        BibliographicReference ref = fetcher.retrieveByHTTP(pubmedIds.get(experiment.getShortName()));
        if (ref == null) {
            if (this.pubmedIdFilename != null) {
                log.warn("Pubmed ID for " + experiment.getShortName() + " was not found in " + this.pubmedIdFilename);
            }
            ref = finder.locatePrimaryReference(experiment);
            if (ref == null) {
                log.error("No ref for " + experiment);
                failedEe.add(experiment.getShortName());
                continue;
            }
        }
        // collect some statistics
        if (experiment.getPrimaryPublication() == null) {
            nullPubCount.add(experiment.getShortName());
        } else if (experiment.getPrimaryPublication().getPubAccession().getAccession().equals(pubmedIds.get(experiment.getShortName()).toString())) {
            samePubCount.add(experiment.getShortName());
        } else {
            diffPubCount.add(experiment.getShortName());
        }
        try {
            log.info("Found pubAccession " + ref.getPubAccession().getAccession() + " for " + experiment);
            ref = (BibliographicReference) ph.persist(ref);
            experiment.setPrimaryPublication(ref);
            ees.update(experiment);
        } catch (Exception e) {
            log.error(experiment.getShortName() + " (id=" + experiment.getId() + ") update failed.");
            e.printStackTrace();
        }
    }
    // print statistics
    log.info("\n\n========== Summary ==========");
    log.info("Total number of experiments: " + expressionExperiments.size());
    log.info("Same publication: " + samePubCount.size());
    log.info("Diff publication: " + diffPubCount.size());
    log.info("No initial publication: " + nullPubCount.size());
    log.info("No publications found: " + failedEe.size());
    log.info("\n\n========== Details ==========");
    log.info("Diff publication: " + Arrays.toString(diffPubCount.toArray()));
    log.info("No initial publication: " + Arrays.toString(nullPubCount.toArray()));
    log.info("No publications found: " + Arrays.toString(failedEe.toArray()));
    return null;
}
Also used : ExpressionExperimentBibRefFinder(ubic.gemma.core.loader.entrez.pubmed.ExpressionExperimentBibRefFinder) BioAssaySet(ubic.gemma.model.expression.experiment.BioAssaySet) ArrayList(java.util.ArrayList) ExpressionExperimentService(ubic.gemma.persistence.service.expression.experiment.ExpressionExperimentService) PubMedXMLFetcher(ubic.gemma.core.loader.entrez.pubmed.PubMedXMLFetcher) Persister(ubic.gemma.persistence.persister.Persister) ExpressionExperiment(ubic.gemma.model.expression.experiment.ExpressionExperiment) BibliographicReference(ubic.gemma.model.common.description.BibliographicReference) IOException(java.io.IOException)

Aggregations

IOException (java.io.IOException)1 ArrayList (java.util.ArrayList)1 ExpressionExperimentBibRefFinder (ubic.gemma.core.loader.entrez.pubmed.ExpressionExperimentBibRefFinder)1 PubMedXMLFetcher (ubic.gemma.core.loader.entrez.pubmed.PubMedXMLFetcher)1 BibliographicReference (ubic.gemma.model.common.description.BibliographicReference)1 BioAssaySet (ubic.gemma.model.expression.experiment.BioAssaySet)1 ExpressionExperiment (ubic.gemma.model.expression.experiment.ExpressionExperiment)1 Persister (ubic.gemma.persistence.persister.Persister)1 ExpressionExperimentService (ubic.gemma.persistence.service.expression.experiment.ExpressionExperimentService)1