Examples with SingleNucleotidePolymorphism uk.ac.ebi.spot.goci.model.SingleNucleotidePolymorphism used on opensource projects

Example 11 with SingleNucleotidePolymorphism

use of uk.ac.ebi.spot.goci.model.SingleNucleotidePolymorphism in project goci by EBISPOT.

the class MappingService method doMapping.

private void doMapping(Association association, String eRelease) throws EnsemblMappingException {
    getLog().info("Mapping association: " + association.getId());
    // Map to store returned location data, this is used in
    // snpLocationMappingService to process all locations linked
    // to a single snp in one go
    Map<String, Set<Location>> snpToLocationsMap = new HashMap<>();
    // Collection to store all genomic contexts
    Collection<GenomicContext> allGenomicContexts = new ArrayList<>();
    // Collection to store all errors for one association
    Collection<String> associationPipelineErrors = new ArrayList<>();
    // For each loci get the SNP and author reported genes
    Collection<Locus> studyAssociationLoci = association.getLoci();
    for (Locus associationLocus : studyAssociationLoci) {
        Long locusId = associationLocus.getId();
        Collection<SingleNucleotidePolymorphism> snpsLinkedToLocus = singleNucleotidePolymorphismQueryService.findByRiskAllelesLociId(locusId);
        Collection<Gene> authorReportedGenesLinkedToSnp = associationLocus.getAuthorReportedGenes();
        // Get gene names
        Collection<String> authorReportedGeneNamesLinkedToSnp = new ArrayList<>();
        for (Gene authorReportedGeneLinkedToSnp : authorReportedGenesLinkedToSnp) {
            authorReportedGeneNamesLinkedToSnp.add(authorReportedGeneLinkedToSnp.getGeneName().trim());
        }
        // Pass rs_id and author reported genes to mapping component
        for (SingleNucleotidePolymorphism snpLinkedToLocus : snpsLinkedToLocus) {
            String snpRsId = snpLinkedToLocus.getRsId();
            EnsemblMappingResult ensemblMappingResult = new EnsemblMappingResult();
            // Try to map supplied data
            try {
                getLog().debug("Running mapping....");
                ensemblMappingResult = ensemblMappingPipeline.run_pipeline(snpRsId, authorReportedGeneNamesLinkedToSnp, eRelease);
            } catch (Exception e) {
                getLog().error("Encountered a " + e.getClass().getSimpleName() + " whilst trying to run mapping of SNP " + snpRsId + ", found in association: " + association.getId(), e);
                throw new EnsemblMappingException();
            }
            getLog().debug("Mapping complete");
            // First remove old locations and genomic contexts
            snpLocationMappingService.removeExistingSnpLocations(snpLinkedToLocus);
            snpGenomicContextMappingService.removeExistingGenomicContexts(snpLinkedToLocus);
            Collection<Location> locations = ensemblMappingResult.getLocations();
            Collection<GenomicContext> snpGenomicContexts = ensemblMappingResult.getGenomicContexts();
            ArrayList<String> pipelineErrors = ensemblMappingResult.getPipelineErrors();
            // Update functional class
            snpLinkedToLocus.setFunctionalClass(ensemblMappingResult.getFunctionalClass());
            snpLinkedToLocus.setLastUpdateDate(new Date());
            snpLinkedToLocus.setMerged(Long.valueOf(ensemblMappingResult.getMerged()));
            // Update the merge table
            if (ensemblMappingResult.getMerged() == 1) {
                String currentSnpId = ensemblMappingResult.getCurrentSnpId();
                SingleNucleotidePolymorphism currentSnp = singleNucleotidePolymorphismRepository.findByRsId(currentSnpId);
                // Add the current SingleNucleotidePolymorphism to the "merged" rsID
                if (currentSnp == null) {
                    currentSnp = new SingleNucleotidePolymorphism();
                    currentSnp.setRsId(currentSnpId);
                    currentSnp.setFunctionalClass(snpLinkedToLocus.getFunctionalClass());
                    singleNucleotidePolymorphismRepository.save(currentSnp);
                    currentSnp = singleNucleotidePolymorphismRepository.findByRsId(currentSnpId);
                }
                snpLinkedToLocus.setCurrentSnp(currentSnp);
            }
            singleNucleotidePolymorphismRepository.save(snpLinkedToLocus);
            // Store location information for SNP
            if (!locations.isEmpty()) {
                for (Location location : locations) {
                    // This would only occur is SNP has multiple locations
                    if (snpToLocationsMap.containsKey(snpRsId)) {
                        snpToLocationsMap.get(snpRsId).add(location);
                    } else // First time we see a SNP store the location
                    {
                        Set<Location> snpLocation = new HashSet<>();
                        snpLocation.add(location);
                        snpToLocationsMap.put(snpRsId, snpLocation);
                    }
                }
            } else {
                getLog().warn("Attempt to map SNP: " + snpRsId + " returned no location details");
                pipelineErrors.add("Attempt to map SNP: " + snpRsId + " returned no location details");
            }
            // Store genomic context data for snp
            if (!snpGenomicContexts.isEmpty()) {
                allGenomicContexts.addAll(snpGenomicContexts);
            } else {
                getLog().warn("Attempt to map SNP: " + snpRsId + " returned no mapped genes");
                pipelineErrors.add("Attempt to map SNP: " + snpRsId + " returned no mapped genes");
            }
            if (!pipelineErrors.isEmpty()) {
                associationPipelineErrors.addAll(pipelineErrors);
            }
        }
    }
    // Create association report based on whether there is errors or not
    if (!associationPipelineErrors.isEmpty()) {
        associationReportService.processAssociationErrors(association, associationPipelineErrors);
    } else {
        associationReportService.updateAssociationReportDetails(association);
    }
    // Save data
    if (!snpToLocationsMap.isEmpty()) {
        getLog().debug("Updating location details ...");
        snpLocationMappingService.storeSnpLocation(snpToLocationsMap);
        getLog().debug("Updating location details complete");
    }
    if (!allGenomicContexts.isEmpty()) {
        getLog().debug("Updating genomic context details ...");
        snpGenomicContextMappingService.processGenomicContext(allGenomicContexts);
        getLog().debug("Updating genomic context details complete");
    }
}

Example 12 with SingleNucleotidePolymorphism

use of uk.ac.ebi.spot.goci.model.SingleNucleotidePolymorphism in project goci by EBISPOT.

the class SingleNucleotidePolymorphismQueryService method findByRsIdIgnoreCase.

@Transactional(readOnly = true)
public SingleNucleotidePolymorphism findByRsIdIgnoreCase(String rsId) {
    SingleNucleotidePolymorphism singleNucleotidePolymorphism = singleNucleotidePolymorphismRepository.findByRsIdIgnoreCase(rsId);
    loadAssociatedData(singleNucleotidePolymorphism);
    return singleNucleotidePolymorphism;
}

Example 13 with SingleNucleotidePolymorphism

use of uk.ac.ebi.spot.goci.model.SingleNucleotidePolymorphism in project goci by EBISPOT.

the class SnpLocationMappingService method storeSnpLocation.

public void storeSnpLocation(Map<String, Set<Location>> snpToLocations) {
    // Go through each rs_id and its associated locations returned from the mapping pipeline
    for (String snpRsId : snpToLocations.keySet()) {
        Set<Location> snpLocationsFromMapping = snpToLocations.get(snpRsId);
        // Check if the SNP exists
        SingleNucleotidePolymorphism snpInDatabase = singleNucleotidePolymorphismQueryService.findByRsIdIgnoreCase(snpRsId);
        if (snpInDatabase != null) {
            // Store all new location objects
            Collection<Location> newSnpLocations = new ArrayList<>();
            for (Location snpLocationFromMapping : snpLocationsFromMapping) {
                String chromosomeNameFromMapping = snpLocationFromMapping.getChromosomeName();
                if (chromosomeNameFromMapping != null) {
                    chromosomeNameFromMapping = chromosomeNameFromMapping.trim();
                }
                Integer chromosomePositionFromMapping = snpLocationFromMapping.getChromosomePosition();
                //                    if (chromosomePositionFromMapping != null) {
                //                        chromosomePositionFromMapping = chromosomePositionFromMapping.trim();
                //                    }
                Region regionFromMapping = snpLocationFromMapping.getRegion();
                String regionNameFromMapping = null;
                if (regionFromMapping != null) {
                    if (regionFromMapping.getName() != null) {
                        regionNameFromMapping = regionFromMapping.getName().trim();
                    }
                }
                // Check if location already exists
                Location existingLocation = locationRepository.findByChromosomeNameAndChromosomePositionAndRegionName(chromosomeNameFromMapping, chromosomePositionFromMapping, regionNameFromMapping);
                if (existingLocation != null) {
                    newSnpLocations.add(existingLocation);
                } else // Create new location
                {
                    Location newLocation = locationCreationService.createLocation(chromosomeNameFromMapping, chromosomePositionFromMapping, regionNameFromMapping);
                    newSnpLocations.add(newLocation);
                }
            }
            // If we have new locations then link to snp and save
            if (newSnpLocations.size() > 0) {
                // Set new location details
                snpInDatabase.setLocations(newSnpLocations);
                // Update the last update date
                snpInDatabase.setLastUpdateDate(new Date());
                singleNucleotidePolymorphismRepository.save(snpInDatabase);
            } else {
                getLog().warn("No new locations to add to " + snpRsId);
            }
        } else // SNP doesn't exist, this should be extremely rare as SNP value is a copy
        // of the variant entered by the curator which
        // by the time mapping is started should already have been saved
        {
            // TODO WHAT WILL HAPPEN FOR MERGED SNPS
            getLog().error("Adding location for SNP not found in database, RS_ID:" + snpRsId);
            throw new RuntimeException("Adding location for SNP not found in database, RS_ID: " + snpRsId);
        }
    }
}

Example 14 with SingleNucleotidePolymorphism

use of uk.ac.ebi.spot.goci.model.SingleNucleotidePolymorphism in project goci by EBISPOT.

the class DefaultGWASOWLPublisher method publishGWASData.

public OWLOntology publishGWASData() throws OWLConversionException {
    // create new ontology
    OWLOntology conversion = getConverter().createConversionOntology();
    // grab all studies from the DAO
    getLog().debug("Fetching studies that require conversion to OWL using StudyRepository...");
    Collection<Study> studies = getStudyService().deepFindPublishedStudies();
    //TODO : check with Tony probably better to do it at the Repository/Service level
    //Discard studies which are not associated with a disease trait and those which haven't been published yet
    //by the GWAS catalog.
    Iterator<Study> iterator = studies.iterator();
    while (iterator.hasNext()) {
        Study study = iterator.next();
        //Remove study which have no diseaseTrait.
        if (study.getDiseaseTrait() == null) {
            iterator.remove();
            getLog().error("Study '" + study.getId() + "' has no disease trait");
        } else if (study.getHousekeeping().getCatalogPublishDate() == null) {
            iterator.remove();
            getLog().error("Study '" + study.getId() + "' has not yet been published");
        } else //Remove studies that have been unpublished
        if (study.getHousekeeping().getCatalogUnpublishDate() != null) {
            iterator.remove();
            getLog().error("Study '" + study.getId() + "' has been unpublished");
        }
    //            }else {
    //
    //                //Remove study which have no associations where pvalue is not null.
    //                Collection<Association> associations = study.getAssociations();
    //                Iterator<Association> associationIterator = associations.iterator();
    //                int associationCount = 0;
    //                while (associationIterator.hasNext()) {
    //                    Association association = associationIterator.next();
    //
    //                    if (association.getPvalueExponent() != null && association.getPvalueMantissa() != null) {
    //                        associationCount++;
    //                    }
    //                }
    //                if (associationCount == 0) {
    //                    iterator.remove();
    //                }
    //            }
    }
    getLog().debug("Query complete, got " + studies.size() + " studies");
    // if studies limit is not set, convert all data, else filter to first n studies and associated data
    if (getStudiesLimit() == -1 && FilterProperties.getDateFilter() == null && FilterProperties.getPvalueFilter() == null) {
        System.out.println("Converting all available data");
        // grab all other data from the DAO
        getLog().debug("Fetching traits that require conversion to OWL using AssociationRepository...");
        Collection<Association> traitAssociations = getAssociationService().findReallyAll();
        //TODO check with Tony how to do that in a better way from service or repository (how to not get associations linked to study with no trait.
        //Discard all the associations which are linked to study which are not linked to a disease trait or haven't
        //been published yet in the GWAS catalog.
        Iterator<Association> associationIterator = traitAssociations.iterator();
        while (associationIterator.hasNext()) {
            Association association = associationIterator.next();
            if (association.getStudy().getDiseaseTrait() == null) {
                associationIterator.remove();
            } else if (association.getStudy().getHousekeeping().getCatalogPublishDate() == null) {
                associationIterator.remove();
            } else if (association.getStudy().getHousekeeping().getCatalogUnpublishDate() != null) {
                associationIterator.remove();
            }
        }
        getLog().debug("Fetching SNPs that require conversion to OWL using SingleNucleotidePolymorphismRepository...");
        Collection<SingleNucleotidePolymorphism> snps = getSingleNucleotidePolymorphismService().findAll();
        getLog().debug("All data fetched");
        // convert this data, starting with SNPs (no dependencies) and working up to studies
        getLog().debug("Starting conversion to OWL...");
        getLog().debug("Converting SNPs...");
        getConverter().addSNPsToOntology(snps, conversion);
        getLog().debug("Converting Trait Associations...");
        getConverter().addAssociationsToOntology(traitAssociations, conversion);
        getLog().debug("Converting Studies...");
        getConverter().addStudiesToOntology(studies, conversion);
        getLog().debug("All conversion done!");
        return conversion;
    } else {
        System.out.println("Data conforming to the filter only");
        return filterAndPublishGWASData(conversion, studies);
    }
}

Example 15 with SingleNucleotidePolymorphism

use of uk.ac.ebi.spot.goci.model.SingleNucleotidePolymorphism in project goci by EBISPOT.

the class AssociationRowProcessorTest method testCreateAssociationFromUploadRow.

@Test
public void testCreateAssociationFromUploadRow() throws Exception {
    // Stubbing mock object behaviour
    when(associationAttributeService.createLocusGenes(STANDARD_ROW.getAuthorReportedGene(), ",")).thenReturn(Arrays.asList(GENE_01, GENE_02));
    when(associationAttributeService.createSnp(STANDARD_ROW.getSnp())).thenReturn(SNP_01);
    when(associationAttributeService.createRiskAllele(STANDARD_ROW.getStrongestAllele(), SNP_01)).thenReturn(RA_01);
    when(associationAttributeService.createSnp(STANDARD_ROW.getProxy())).thenReturn(PROXY);
    Association association = associationRowProcessor.createAssociationFromUploadRow(STANDARD_ROW);
    verify(associationCalculationService, never()).reverseCI(STANDARD_ROW.getRange());
    verify(associationCalculationService, never()).setRange(STANDARD_ROW.getStandardError(), STANDARD_ROW.getOrPerCopyNum());
    verify(associationAttributeService, never()).getEfoTraitsFromRepository(Collections.EMPTY_LIST);
    verify(associationAttributeService, times(1)).createLocusGenes(STANDARD_ROW.getAuthorReportedGene(), ",");
    verify(associationAttributeService, times(1)).createSnp(STANDARD_ROW.getSnp());
    verify(associationAttributeService, times(1)).createSnp(STANDARD_ROW.getProxy());
    verify(associationAttributeService, times(1)).createRiskAllele(STANDARD_ROW.getStrongestAllele(), SNP_01);
    assertThat(association).extracting("id", "riskFrequency", "pvalueDescription", "pvalueMantissa", "pvalueExponent", "multiSnpHaplotype", "snpInteraction", "snpApproved", "snpType", "standardError", "range", "description", "orPerCopyNum", "orPerCopyRecip", "orPerCopyRecipRange", "betaNum", "betaUnit", "betaDirection", "study", "associationReport", "lastMappingDate", "lastMappingPerformedBy", "lastUpdateDate").containsExactly(null, "0.52", "(some pvalue description)", 2, -7, false, false, false, null, (float) 0.6, "[0.82-0.92]", null, (float) 1.22, null, null, null, null, null, null, null, null, null, null);
    assertThat(association.getEfoTraits()).isEmpty();
    assertThat(association.getEvents()).isEmpty();
    assertThat(association.getStudy()).isNull();
    assertThat(association.getLoci()).hasSize(1);
    // Check locus attributes
    Collection<Gene> locusGenes = new ArrayList<>();
    association.getLoci().stream().forEach(locus -> {
        locusGenes.addAll(locus.getAuthorReportedGenes());
    });
    Collection<RiskAllele> locusRiskAlleles = new ArrayList<>();
    association.getLoci().stream().forEach(locus -> {
        locusRiskAlleles.addAll(locus.getStrongestRiskAlleles());
    });
    Collection<SingleNucleotidePolymorphism> proxies = new ArrayList<>();
    locusRiskAlleles.stream().forEach(riskAllele -> {
        proxies.addAll(riskAllele.getProxySnps());
    });
    assertThat(association.getLoci()).extracting(Locus::getDescription).containsOnly("Single variant");
    assertThat(locusGenes).hasSize(2).contains(GENE_01, GENE_02);
    assertThat(locusRiskAlleles).hasSize(1).contains(RA_01);
    assertThat(locusRiskAlleles).extracting("riskAlleleName", "riskFrequency", "snp.rsId").contains(tuple("rs123-?", "0.52", "rs123"));
    assertThat(locusRiskAlleles).extracting(RiskAllele::getSnp).contains(SNP_01);
    assertThat(proxies).contains(PROXY);
    assertThat(proxies).extracting(SingleNucleotidePolymorphism::getRsId).containsExactly("rs99");
}