Examples with HotSpotScoring de.bioforscher.jstructure.efr.model.HotSpotScoring used on opensource projects

Example 1 with HotSpotScoring

use of de.bioforscher.jstructure.efr.model.HotSpotScoring in project jstructure by JonStargaryen.

the class A06_WriteStructuralInformationByResidueCsv 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());
        // boolean sane = split[6].equalsIgnoreCase("true");
        Structure structure = StructureParser.fromPdbId(pdbId).parse();
        Chain chain = structure.chains().findFirst().get();
        Path start2foldXml = Start2FoldConstants.XML_DIRECTORY.resolve(entryId + ".xml");
        Start2FoldXmlParser.parseStability(chain, start2foldXml);
        Start2FoldXmlParser.parseSpecificExperiment(chain, start2foldXml, experimentIds);
        try {
            EvolutionaryCouplingParser.parseHotSpotFile(chain, Start2FoldConstants.COUPLING_DIRECTORY.resolve(entryId.toUpperCase() + "_hs.html"));
        } catch (Exception e) {
        }
        boolean ecAnnotation = chain.aminoAcids().anyMatch(residue -> residue.getFeature(HotSpotScoring.class).getEcCount() > 0);
        List<AminoAcid> topScoringResidues;
        if (ecAnnotation) {
            double fraction = 0.4;
            int residuesToSelect = (int) (fraction * chain.aminoAcids().count());
            topScoringResidues = chain.aminoAcids().sorted(Comparator.comparingDouble((AminoAcid aminoAcid) -> aminoAcid.getFeature(HotSpotScoring.class).getCumStrength()).reversed()).limit(residuesToSelect).collect(Collectors.toList());
        } else {
            topScoringResidues = new ArrayList<>();
        }
        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<AminoAcid> strongResidues = chain.aminoAcids().filter(aminoAcid -> aminoAcid.getFeature(Start2FoldResidueAnnotation.class).isStrong()).collect(Collectors.toList());
        List<Integer> functionalResidueNumbers = Start2FoldConstants.extractFunctionalResidueNumbers(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());
        List<ContactStructuralInformation> contactStructuralInformation = StructuralInformationParserService.getInstance().parseContactStructuralInformation(Start2FoldConstants.DATA_DIRECTORY.resolve("si").resolve("raw").resolve(entryId.toUpperCase() + ".out"), chain, earlyFoldingResidues);
        List<ResidueStructuralInformation> residueStructuralInformation = StructuralInformationParserService.getInstance().composeResidueStructuralInformation(aminoAcids, earlyFoldingResidues, contactStructuralInformation);
        ResidueGraph conventionalProteinGraph = ResidueGraph.createResidueGraph(chain, ContactDefinitionFactory.createAlphaCarbonContactDefinition(8.0));
        List<AminoAcid> residuesInEarlyFoldingSecondaryStructureElements = chain.aminoAcids().filter(aminoAcid -> !aminoAcid.getFeature(GenericSecondaryStructure.class).getSecondaryStructure().isCoilType()).filter(aminoAcid -> {
            GenericSecondaryStructure.SecondaryStructureElement surroundingSecondaryStructureElement = aminoAcid.getFeature(GenericSecondaryStructure.class).getSurroundingSecondaryStructureElement(aminoAcid);
            List<AminoAcid> surroundingAminoAcids = chain.getAminoAcids().subList(surroundingSecondaryStructureElement.getStart(), surroundingSecondaryStructureElement.getEnd() + 1);
            return surroundingAminoAcids.stream().anyMatch(earlyFoldingResidues::contains);
        }).collect(Collectors.toList());
        System.out.println("efr: " + (earlyFoldingResidues.size() > 0) + " strong: " + (strongResidues.size() > 0) + " functional: " + (functionalResidues.size() > 0));
        return Optional.of(chain.aminoAcids().filter(AminoAcid::isStandardAminoAcid).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();
            ResidueStructuralInformation residueStructuralInformationEntry = residueStructuralInformation.get(aminoAcid.getAminoAcidIndex());
            GenericSecondaryStructure.SecondaryStructureElement surroundingSecondaryStructureElement = sse.getSurroundingSecondaryStructureElement(aminoAcid);
            int sseSize = surroundingSecondaryStructureElement.getSize();
            if (sse.getSecondaryStructure().isCoilType()) {
                sseSize = 0;
            }
            int sseTerminusDistance = surroundingSecondaryStructureElement.getTerminusDistance();
            if (sse.getSecondaryStructure().isCoilType()) {
                sseTerminusDistance = -1;
            }
            return pdbId + "," + "A" + "," + aminoAcid.getResidueIdentifier() + "," + aminoAcid.getOneLetterCode() + "," + sse.getSecondaryStructure().getReducedRepresentation() + "," + sse.getSecondaryStructure().getOneLetterRepresentation() + "," + (sse.getSecondaryStructure().isHelixType() ? "true" : "false") + "," + (sse.getSecondaryStructure().isStrandType() ? "true" : "false") + "," + (sse.getSecondaryStructure().isCoilType() ? "true" : "false") + "," + sseSize + "," + sseTerminusDistance + "," + (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() + "," + topScoringResidues.contains(aminoAcid) + "," + 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()) + "," + StandardFormat.format(residueStructuralInformationEntry.getAverageRmsdIncrease()) + "," + StandardFormat.format(residueStructuralInformationEntry.getAverageTmScoreIncrease()) + "," + StandardFormat.format(residueStructuralInformationEntry.getAverageQIncrease()) + "," + StandardFormat.format(residueStructuralInformationEntry.getMaximumRmsdIncrease()) + "," + StandardFormat.format(residueStructuralInformationEntry.getMaximumTmScoreIncrease()) + "," + StandardFormat.format(residueStructuralInformationEntry.getMaximumQIncrease()) + "," + StandardFormat.format(residueStructuralInformationEntry.getAverageRmsdIncreaseZScore()) + "," + StandardFormat.format(residueStructuralInformationEntry.getFractionOfTopScoringContacts()) + "," + (earlyFoldingResidues.contains(aminoAcid) ? "early" : "late") + "," + (residuesInEarlyFoldingSecondaryStructureElements.contains(aminoAcid) ? "true" : "false") + "," + functionalAnnotation + "," + (strongResidues.contains(aminoAcid) ? "strong" : "weak") + "," + (earlyFoldingResidues.size() > 0) + "," + (strongResidues.size() > 0) + "," + (functionalResidues.size() > 0) + "," + ecAnnotation;
        }).collect(Collectors.joining(System.lineSeparator())));
    } catch (Exception e) {
        logger.info("calculation failed for {}\nby: {}", line, e.getMessage());
        return Optional.empty();
    }
}

Example 2 with HotSpotScoring

use of de.bioforscher.jstructure.efr.model.HotSpotScoring in project jstructure by JonStargaryen.

the class A02_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.parseStability(chain, Start2FoldConstants.XML_DIRECTORY.resolve(entryId + ".xml"));
        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<AminoAcid> stableResidues = chain.aminoAcids().filter(aminoAcid -> aminoAcid.getFeature(Start2FoldResidueAnnotation.class).isStrong()).collect(Collectors.toList());
        List<Integer> functionalResidueNumbers = Start2FoldConstants.extractFunctionalResidueNumbers(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());
        ResidueGraph conventionalProteinGraph = ResidueGraph.createResidueGraph(chain, ContactDefinitionFactory.createAlphaCarbonContactDefinition(8));
        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 + "," + (stableResidues.contains(aminoAcid) ? "stable" : "unstable");
        }).collect(Collectors.joining(System.lineSeparator())));
    } catch (Exception e) {
        logger.info("calculation failed for {}", line, e);
        return Optional.empty();
    }
}

Example 3 with HotSpotScoring

use of de.bioforscher.jstructure.efr.model.HotSpotScoring in project jstructure by JonStargaryen.

the class EvolutionaryCouplingParser method parseHotSpotFile.

public 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);
        // System.out.println(aminoAcid.getOneLetterCode() + " " + aa);
        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()));
}