Examples with AminoAcid de.bioforscher.jstructure.model.structure.aminoacid.AminoAcid used on opensource projects

Example 36 with AminoAcid

use of de.bioforscher.jstructure.model.structure.aminoacid.AminoAcid in project jstructure by JonStargaryen.

the class StructureCollectorsTest method shouldCreateDistinctContainer.

@Test
public void shouldCreateDistinctContainer() {
    // create copy
    AminoAcid original = protein.select().aminoAcids().asAminoAcid();
    String initialPdbRecord = original.getPdbRepresentation();
    AtomContainer copy = original.createDeepCopy();
    // manipulate coordinates of copy
    copy.getAtoms().get(0).setCoordinates(new double[] { 0, 0, 0 });
    // should not affect original
    Assert.assertNotEquals(initialPdbRecord, copy.getPdbRepresentation());
    Assert.assertEquals(initialPdbRecord, original.getPdbRepresentation());
}

Example 37 with AminoAcid

use of de.bioforscher.jstructure.model.structure.aminoacid.AminoAcid in project jstructure by JonStargaryen.

the class StructureParserTest method shouldHandleProteinWithOligopeptide.

@Test
public void shouldHandleProteinWithOligopeptide() {
    Structure protein = StructureParser.fromPdbId("1lyb").parse();
    // contains peptide-like groups - ought to be an AminoAcid
    AminoAcid statine = (AminoAcid) protein.select().groupName("STA").asGroup();
    Assert.assertTrue(statine.getPolymerType() == GroupPrototype.PolymerType.PEPTIDE_LIKE);
    Assert.assertTrue(statine.isAminoAcid());
    Assert.assertFalse(statine.isLigand());
    Assert.assertTrue(protein.aminoAcids().collect(Collectors.toList()).contains(statine));
}

Example 38 with AminoAcid

use of de.bioforscher.jstructure.model.structure.aminoacid.AminoAcid in project jstructure by JonStargaryen.

the class A03_WriteDatasetCsv method handleLine.

private static Optional<String> handleLine(String line) {
    try {
        System.out.println(line);
        String[] split = line.split(";");
        String entryId = split[0];
        String pdbId = split[1];
        List<Integer> experimentIds = Pattern.compile(",").splitAsStream(split[2].replaceAll("\\[", "").replaceAll("]", "")).map(Integer::valueOf).collect(Collectors.toList());
        Structure structure = StructureParser.fromPdbId(pdbId).parse();
        Chain chain = structure.chains().findFirst().get();
        Start2FoldXmlParser.parseSpecificExperiment(chain, Start2FoldConstants.XML_DIRECTORY.resolve(entryId + ".xml"), experimentIds);
        EvolutionaryCouplingParser.parseHotSpotFile(chain, Start2FoldConstants.COUPLING_DIRECTORY.resolve(entryId.toUpperCase() + "_hs.html"));
        EQuantParser.parseEQuantFile(chain, Start2FoldConstants.EQUANT_DIRECTORY.resolve(entryId.toLowerCase() + ".equant-small.txt"));
        List<AminoAcid> earlyFoldingResidues = chain.aminoAcids().filter(aminoAcid -> aminoAcid.getFeature(Start2FoldResidueAnnotation.class).isEarly()).collect(Collectors.toList());
        List<Integer> functionalResidueNumbers = Start2FoldConstants.extractFunctioanlResidueNumbers(split);
        List<AminoAcid> functionalResidues = new ArrayList<>();
        // do nothing if no annotation of functional residues exists
        if (!functionalResidueNumbers.isEmpty()) {
            FunctionalResidueParser.parse(chain, functionalResidueNumbers);
            chain.aminoAcids().filter(aminoAcid -> aminoAcid.getFeature(FunctionalResidueAnnotation.class).isFunctional()).forEach(functionalResidues::add);
        }
        List<AminoAcid> aminoAcids = chain.aminoAcids().collect(Collectors.toList());
        ProteinGraph conventionalProteinGraph = ProteinGraphFactory.createProteinGraph(chain, ProteinGraphFactory.InteractionScheme.CALPHA8);
        return Optional.of(chain.aminoAcids().map(aminoAcid -> {
            GenericSecondaryStructure sse = aminoAcid.getFeature(GenericSecondaryStructure.class);
            HotSpotScoring hotSpotScoring = aminoAcid.getFeature(HotSpotScoring.class);
            PLIPInteractionContainer plipInteractionContainer = aminoAcid.getFeature(PLIPInteractionContainer.class);
            PLIPInteractionContainer nonLocalPlipInteractionContainer = new PLIPInteractionContainer(null, plipInteractionContainer.getInteractions().stream().filter(inter -> Math.abs(inter.getPartner1().getResidueIndex() - inter.getPartner2().getResidueIndex()) > 5).collect(Collectors.toList()));
            PLIPInteractionContainer localPlipInteractionContainer = new PLIPInteractionContainer(null, plipInteractionContainer.getInteractions().stream().filter(inter -> !nonLocalPlipInteractionContainer.getInteractions().contains(inter)).collect(Collectors.toList()));
            String equantScore = "NA";
            try {
                equantScore = StandardFormat.format(aminoAcid.getFeature(EQuantScore.class).getEvaluation());
            } catch (ComputationException e) {
                logger.warn("missing equant scoring for {}", aminoAcid);
            }
            String functionalAnnotation = "NA";
            if (functionalResidues.size() > 0) {
                functionalAnnotation = functionalResidues.contains(aminoAcid) ? "functional" : "non-functional";
            }
            ResidueTopologicPropertiesContainer residueTopologicPropertiesContainer = aminoAcid.getFeature(ResidueTopologicPropertiesContainer.class);
            double terminusDistance = aminoAcids.indexOf(aminoAcid);
            terminusDistance = Math.min(terminusDistance, aminoAcids.size() - terminusDistance);
            terminusDistance /= (double) aminoAcids.size();
            return pdbId + "," + "A" + "," + aminoAcid.getResidueIdentifier() + "," + aminoAcid.getOneLetterCode() + "," + sse.getSecondaryStructure().getReducedRepresentation() + "," + sse.getSecondaryStructure().getOneLetterRepresentation() + "," + sse.getSurroundingSecondaryStructureElement(aminoAcid).getSize() + "," + (aminoAcid.getFeature(AccessibleSurfaceArea.class).isExposed() ? "exposed" : "buried") + "," + StandardFormat.format(aminoAcid.getFeature(GeometricProperties.class).getDistanceToCentroid()) + "," + StandardFormat.format(terminusDistance) + "," + plipInteractionContainer.getHydrogenBonds().size() + "," + plipInteractionContainer.getHydrophobicInteractions().size() + "," + plipInteractionContainer.getBackboneInteractions().size() + "," + plipInteractionContainer.getInteractions().size() + "," + localPlipInteractionContainer.getHydrogenBonds().size() + "," + localPlipInteractionContainer.getHydrophobicInteractions().size() + "," + localPlipInteractionContainer.getBackboneInteractions().size() + "," + localPlipInteractionContainer.getInteractions().size() + "," + nonLocalPlipInteractionContainer.getHydrogenBonds().size() + "," + nonLocalPlipInteractionContainer.getHydrophobicInteractions().size() + "," + nonLocalPlipInteractionContainer.getBackboneInteractions().size() + "," + nonLocalPlipInteractionContainer.getInteractions().size() + "," + StandardFormat.format(aminoAcid.getFeature(EnergyProfile.class).getSolvationEnergy()) + "," + StandardFormat.format(aminoAcid.getFeature(EgorAgreement.class).getEgorPrediction()) + "," + equantScore + "," + StandardFormat.format(aminoAcid.getFeature(AccessibleSurfaceArea.class).getRelativeAccessibleSurfaceArea()) + "," + StandardFormat.format(aminoAcid.getFeature(LoopFraction.class).getLoopFraction()) + "," + hotSpotScoring.getEcCount() + "," + StandardFormat.format(hotSpotScoring.getCumStrength()) + "," + StandardFormat.format(hotSpotScoring.getEcStrength()) + "," + hotSpotScoring.getConservation() + "," + StandardFormat.format(residueTopologicPropertiesContainer.getFullPlip().getBetweenness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getFullPlip().getCloseness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getFullPlip().getClusteringCoefficient()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getHydrogenPlip().getBetweenness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getHydrogenPlip().getCloseness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getHydrogenPlip().getClusteringCoefficient()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getHydrophobicPlip().getBetweenness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getHydrophobicPlip().getCloseness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getHydrophobicPlip().getClusteringCoefficient()) + "," + conventionalProteinGraph.getContactsOf(aminoAcid).size() + "," + conventionalProteinGraph.getLocalContactsOf(aminoAcid).size() + "," + conventionalProteinGraph.getNonLocalContactsOf(aminoAcid).size() + "," + StandardFormat.format(residueTopologicPropertiesContainer.getConventional().getBetweenness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getConventional().getCloseness()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getConventional().getClusteringCoefficient()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getFullPlip().getDistinctNeighborhoodCount()) + "," + StandardFormat.format(residueTopologicPropertiesContainer.getConventional().getDistinctNeighborhoodCount()) + "," + (earlyFoldingResidues.contains(aminoAcid) ? "early" : "late") + "," + functionalAnnotation;
        }).collect(Collectors.joining(System.lineSeparator())));
    } catch (Exception e) {
        logger.info("calculation failed for {}", line, e);
        return Optional.empty();
    }
}

Example 39 with AminoAcid

use of de.bioforscher.jstructure.model.structure.aminoacid.AminoAcid in project jstructure by JonStargaryen.

the class A07_WriteCouplingRangeCsv method handleLine.

private static Optional<String> handleLine(String line) {
    try {
        System.out.println(line);
        String[] split = line.split(";");
        String entryId = split[0];
        String pdbId = split[1];
        List<Integer> experimentIds = Pattern.compile(",").splitAsStream(split[2].replaceAll("\\[", "").replaceAll("]", "")).map(Integer::valueOf).collect(Collectors.toList());
        Structure structure = StructureParser.fromPdbId(pdbId).parse();
        Chain chain = structure.chains().findFirst().get();
        Start2FoldXmlParser.parseSpecificExperiment(chain, Start2FoldConstants.XML_DIRECTORY.resolve(entryId + ".xml"), experimentIds);
        EvolutionaryCouplingParser.parseHotSpotFile(chain, Start2FoldConstants.COUPLING_DIRECTORY.resolve(entryId.toUpperCase() + "_hs.html"));
        List<AminoAcid> earlyFoldingResidues = chain.aminoAcids().filter(aminoAcid -> aminoAcid.getFeature(Start2FoldResidueAnnotation.class).isEarly()).collect(Collectors.toList());
        Map<Integer, List<Double>> localPlmScores = new HashMap<>();
        Map<Integer, List<Double>> longRangePlmScores = new HashMap<>();
        Document hotSpotDocument = Jsoup.parse(Files.readAllLines(Paths.get("/home/bittrich/git/phd_sb_repo/data/start2fold/coupling/" + entryId + "_ec.html")).stream().collect(Collectors.joining(System.lineSeparator())));
        List<AminoAcid> aminoAcids = chain.aminoAcids().collect(Collectors.toList());
        for (int i = 0; i < aminoAcids.size(); i++) {
            localPlmScores.put(i, new ArrayList<>());
            longRangePlmScores.put(i, new ArrayList<>());
        }
        hotSpotDocument.getElementsByTag("tr").stream().skip(1).forEach(element -> {
            Elements tds = element.getElementsByTag("td");
            int residueNumber1 = Integer.valueOf(tds.get(2).text()) - 1;
            int residueNumber2 = Integer.valueOf(tds.get(4).text()) - 1;
            double plmScore = Double.valueOf(tds.get(6).text());
            boolean localContact = Math.abs(residueNumber1 - residueNumber2) < 6;
            if (localContact) {
                System.out.println("local contact: " + element.text());
                localPlmScores.get(residueNumber1).add(plmScore);
                localPlmScores.get(residueNumber2).add(plmScore);
            } else {
                System.out.println("long-range contact: " + element.text());
                longRangePlmScores.get(residueNumber1).add(plmScore);
                longRangePlmScores.get(residueNumber2).add(plmScore);
            }
        });
        return Optional.of(aminoAcids.stream().map(aminoAcid -> pdbId + ",A," + aminoAcid.getOneLetterCode() + "," + aminoAcid.getResidueIdentifier().getResidueNumber() + "," + (earlyFoldingResidues.contains(aminoAcid) ? "early" : "late") + "," + "local," + StandardFormat.format(localPlmScores.get(aminoAcid.getResidueIndex()).stream().mapToDouble(Double::valueOf).average().orElse(0.0)) + System.lineSeparator() + pdbId + ",A," + aminoAcid.getOneLetterCode() + "," + aminoAcid.getResidueIdentifier().getResidueNumber() + "," + (earlyFoldingResidues.contains(aminoAcid) ? "early" : "late") + "," + "long-range," + StandardFormat.format(longRangePlmScores.get(aminoAcid.getResidueIndex()).stream().mapToDouble(Double::valueOf).average().orElse(0.0))).collect(Collectors.joining(System.lineSeparator())));
    } catch (Exception e) {
        e.printStackTrace();
        return Optional.empty();
    }
}

Example 40 with AminoAcid

use of de.bioforscher.jstructure.model.structure.aminoacid.AminoAcid in project jstructure by JonStargaryen.

the class EvolutionaryCouplingParser method parseHotSpotFile.

static void parseHotSpotFile(Chain chain, InputStream inputStream) {
    Document hotSpotDocument = Jsoup.parse(new BufferedReader(new InputStreamReader(inputStream)).lines().collect(Collectors.joining(System.lineSeparator())));
    List<AminoAcid> aminoAcids = chain.aminoAcids().collect(Collectors.toList());
    hotSpotDocument.getElementsByTag("tr").stream().skip(1).forEach(element -> {
        Elements tds = element.getElementsByTag("td");
        int residueNumber = Integer.valueOf(tds.get(0).text());
        String aa = tds.get(1).text();
        int ecCount = Integer.valueOf(tds.get(2).text());
        double cumStrength = Double.valueOf(tds.get(3).text());
        double ecStrength = Double.valueOf(tds.get(4).text());
        int conservation = Integer.valueOf(tds.get(5).text());
        AminoAcid aminoAcid = aminoAcids.get(residueNumber - 1);
        aminoAcid.getFeatureContainer().addFeature(new HotSpotScoring(ecCount, cumStrength, ecStrength, conservation));
    });
    // assign baseline
    aminoAcids.stream().map(AminoAcid::getFeatureContainer).filter(featureContainer -> !featureContainer.getFeatureOptional(HotSpotScoring.class).isPresent()).forEach(featureContainer -> featureContainer.addFeature(new HotSpotScoring()));
}