Search in sources :

Example 1 with ReferencePoints

use of io.repseq.core.ReferencePoints in project repseqio by repseqio.

the class DebugAction method go.

@Override
public void go(ActionHelper helper) throws Exception {
    VDJCLibraryRegistry reg = VDJCLibraryRegistry.getDefault();
    reg.registerLibraries(params.getInput());
    Pattern namePattern = params.name == null ? null : Pattern.compile(params.name);
    GeneFeature cdr3FirstTriplet = new GeneFeature(ReferencePoint.CDR3Begin, 0, 3);
    GeneFeature cdr3LastTriplet = new GeneFeature(ReferencePoint.CDR3End, -3, 0);
    GeneFeature vIntronDonor = new GeneFeature(ReferencePoint.VIntronBegin, 0, 2);
    GeneFeature vIntronAcceptor = new GeneFeature(ReferencePoint.VIntronEnd, -2, 0);
    for (VDJCLibrary lib : reg.getLoadedLibraries()) {
        for (VDJCGene gene : lib.getGenes()) {
            if (namePattern != null && !namePattern.matcher(gene.getName()).matches())
                continue;
            // First generate list of warning messages
            List<String> warnings = new ArrayList<>();
            if (gene.isFunctional() || params.getCheckAll()) {
                NucleotideSequence l3;
                switch(gene.getGeneType()) {
                    case Variable:
                        // Flag AA residues flanking CDR3
                        l3 = gene.getFeature(cdr3FirstTriplet);
                        if (l3 == null)
                            warnings.add("Unable to find CDR3 start");
                        else if (l3.size() != 3)
                            warnings.add("Unable to translate sequence: " + gene.getName() + " / " + l3);
                        else if (AminoAcidSequence.translate(l3).codeAt(0) != AminoAcidAlphabet.C)
                            warnings.add("CDR3 does not start with C, was: " + l3.toString() + " / " + AminoAcidSequence.translate(l3).toString() + " / CDR3Begin: " + gene.getData().getAnchorPoints().get(ReferencePoint.CDR3Begin));
                        // Flag suspicious exon borders
                        // https://schneider.ncifcrf.gov/gallery/SequenceLogoSculpture.gif
                        NucleotideSequence vIntronDonorSeq = gene.getFeature(vIntronDonor);
                        if (vIntronDonorSeq != null && !vIntronDonorSeq.toString().equals("GT") && !vIntronDonorSeq.toString().equals("GC"))
                            warnings.add("Expected VIntron sequence to start with GT, was: " + vIntronDonorSeq.toString());
                        NucleotideSequence vIntronAcceptorSeq = gene.getFeature(vIntronAcceptor);
                        if (vIntronAcceptorSeq != null && !vIntronAcceptorSeq.toString().equals("AG"))
                            warnings.add("Expected VIntron sequence to end with AG, was: " + vIntronAcceptorSeq.toString());
                        ReferencePoints partitioning = gene.getPartitioning();
                        if (partitioning.isAvailable(GeneFeature.VTranscriptWithout5UTR)) {
                            // Iterating over all reading-frame bound anchor points of V gene
                            for (ReferencePoint anchorPoint : ReferencePoint.DefaultReferencePoints) {
                                if (anchorPoint.getGeneType() != GeneType.Variable || !anchorPoint.isTripletBoundary())
                                    continue;
                                // And checking that they are in the same frame as Start (L1Begin)
                                int relativePosition = partitioning.getRelativePosition(GeneFeature.VTranscriptWithout5UTR, anchorPoint);
                                if (// Point is defined
                                relativePosition >= 0 && relativePosition % 3 != 0)
                                    warnings.add("Expected " + anchorPoint + " to have position dividable by three inside VTranscriptWithout5UTR. " + "This may indicate an error in the L2 boundaries.");
                            }
                        }
                        break;
                    case Joining:
                        // Flag AA residues flanking CDR3
                        l3 = gene.getFeature(cdr3LastTriplet);
                        if (l3 == null)
                            warnings.add("Unable to find CDR3 end");
                        else if (l3.size() != 3)
                            warnings.add("Unable to translate sequence: " + gene.getName() + " / " + l3);
                        else if (AminoAcidSequence.translate(l3).codeAt(0) != AminoAcidAlphabet.W && AminoAcidSequence.translate(l3).codeAt(0) != AminoAcidAlphabet.F)
                            warnings.add("CDR3 does not end with W or F, was: " + l3.toString() + " / " + AminoAcidSequence.translate(l3).toString() + " / CDR3End: " + gene.getData().getAnchorPoints().get(ReferencePoint.CDR3End));
                        break;
                }
                for (GeneFeature geneFeature : aaGeneFeatures.get(gene.getGeneType())) {
                    AminoAcidSequence aaSequence = getAminoAcidSequence(gene, geneFeature, gene.getFeature(geneFeature));
                    if (aaSequence != null) {
                        // Flag if contains stop codon
                        if (aaSequence.numberOfStops() > 0)
                            warnings.add(GeneFeature.encode(geneFeature) + " contains a stop codon");
                    }
                }
            }
            if (params.getProblemOnly() && warnings.isEmpty())
                continue;
            System.out.println(gene.getName() + " (" + (gene.isFunctional() ? "F" : "P") + ") " + gene.getChains() + " : " + lib.getTaxonId());
            if (!warnings.isEmpty()) {
                System.out.println();
                System.out.println("WARNINGS: ");
                for (String warning : warnings) {
                    System.out.println(warning);
                }
                System.out.println();
            }
            for (GeneFeature geneFeature : geneFeatures.get(gene.getGeneType())) {
                System.out.println();
                System.out.println(GeneFeature.encode(geneFeature));
                NucleotideSequence nSequence = gene.getFeature(geneFeature);
                AminoAcidSequence aaSequence = getAminoAcidSequence(gene, geneFeature, nSequence);
                System.out.print("N:   ");
                if (nSequence == null)
                    System.out.println("Not Available");
                else
                    System.out.println(nSequence);
                if (GeneFeature.getFrameReference(geneFeature) != null) {
                    System.out.print("AA:  ");
                    if (aaSequence == null)
                        System.out.println("Not Available");
                    else
                        System.out.println(aaSequence);
                }
            }
            System.out.println("=========");
            System.out.println();
        }
    }
}
Also used : Pattern(java.util.regex.Pattern) GeneFeature(io.repseq.core.GeneFeature) ArrayList(java.util.ArrayList) ReferencePoints(io.repseq.core.ReferencePoints) ReferencePoint(io.repseq.core.ReferencePoint) ReferencePoint(io.repseq.core.ReferencePoint) AminoAcidSequence(com.milaboratory.core.sequence.AminoAcidSequence) NucleotideSequence(com.milaboratory.core.sequence.NucleotideSequence) VDJCLibrary(io.repseq.core.VDJCLibrary) VDJCGene(io.repseq.core.VDJCGene) VDJCLibraryRegistry(io.repseq.core.VDJCLibraryRegistry)

Example 2 with ReferencePoints

use of io.repseq.core.ReferencePoints in project repseqio by repseqio.

the class DebugAction method getAminoAcidSequence.

private static AminoAcidSequence getAminoAcidSequence(VDJCGene gene, GeneFeature geneFeature, NucleotideSequence nSequence) {
    ReferencePoints partitioning = gene.getPartitioning();
    ReferencePoint frameReference = GeneFeature.getFrameReference(geneFeature);
    AminoAcidSequence aaSequence;
    if (frameReference != null) {
        int relativePosition = partitioning.getRelativePosition(geneFeature, frameReference);
        aaSequence = nSequence == null || relativePosition < 0 ? null : AminoAcidSequence.translate(nSequence, withIncompleteCodon(relativePosition));
    } else
        aaSequence = null;
    return aaSequence;
}
Also used : ReferencePoint(io.repseq.core.ReferencePoint) AminoAcidSequence(com.milaboratory.core.sequence.AminoAcidSequence) ReferencePoints(io.repseq.core.ReferencePoints) ReferencePoint(io.repseq.core.ReferencePoint)

Aggregations

AminoAcidSequence (com.milaboratory.core.sequence.AminoAcidSequence)2 ReferencePoint (io.repseq.core.ReferencePoint)2 ReferencePoints (io.repseq.core.ReferencePoints)2 NucleotideSequence (com.milaboratory.core.sequence.NucleotideSequence)1 GeneFeature (io.repseq.core.GeneFeature)1 VDJCGene (io.repseq.core.VDJCGene)1 VDJCLibrary (io.repseq.core.VDJCLibrary)1 VDJCLibraryRegistry (io.repseq.core.VDJCLibraryRegistry)1 ArrayList (java.util.ArrayList)1 Pattern (java.util.regex.Pattern)1