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

Example 1 with ContactStructuralInformation

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

the class A03_ReconstructByVariousStrategy method handleChain.

private static void handleChain(ExplorerChain explorerChain) {
    logger.info("[{}] starting job", explorerChain.getStfId());
    try {
        Chain nativeChain = explorerChain.getChain();
        Path nativeChainPath = Files.createTempFile("nativechain-", ".pdb");
        Files.write(nativeChainPath, nativeChain.getPdbRepresentation().getBytes());
        List<ContactStructuralInformation> contactStructuralInformation = explorerChain.getContacts();
        // annotate with PLIP data
        PLIPInteractionContainer plipInteractionContainer = nativeChain.getFeature(PLIPInteractionContainer.class);
        for (ContactStructuralInformation csi : contactStructuralInformation) {
            AminoAcid aminoAcid1 = nativeChain.select().residueNumber(csi.getResidueIdentifier1()).asAminoAcid();
            AminoAcid aminoAcid2 = nativeChain.select().residueNumber(csi.getResidueIdentifier2()).asAminoAcid();
            if (plipInteractionContainer.getHydrogenBonds().stream().anyMatch(hydrogenBond -> isContact(hydrogenBond, aminoAcid1, aminoAcid2))) {
                csi.markAsHydrogenBond();
            }
            if (plipInteractionContainer.getHydrophobicInteractions().stream().anyMatch(hydrophobicInteraction -> isContact(hydrophobicInteraction, aminoAcid1, aminoAcid2))) {
                csi.markAsHydrophobicInteraction();
            }
        }
        int numberOfNativeContacts = contactStructuralInformation.size();
        int numberOfContactsToSelect = (int) (numberOfNativeContacts * DEFAULT_COVERAGE);
        List<ReconstructionContactMap> contactMaps = Stream.of(ReconstructionStrategyDefinition.values()).map(ReconstructionStrategyDefinition::getReconstructionStrategy).flatMap(reconstructionStrategy -> IntStream.range(0, REDUNDANCY).boxed().flatMap(i -> {
            if (!reconstructionStrategy.isNegatable()) {
                ReconstructionContactMap contactMap = reconstructionStrategy.composeReconstructionContactMap(nativeChain, contactStructuralInformation, numberOfContactsToSelect);
                contactMap.setName(reconstructionStrategy.getName() + "-" + (i + 1));
                return Stream.of(contactMap);
            } else {
                // short, long, hydrogen, and hydrophobic bins have to be negated explicitly to get comparable results
                Pair<ReconstructionContactMap, ReconstructionContactMap> contactMapPair = reconstructionStrategy.composeReconstructionAndNegatedReconstructionContactMap(nativeChain, contactStructuralInformation, numberOfContactsToSelect);
                contactMapPair.getLeft().setName(reconstructionStrategy.getName() + "-" + (i + 1));
                contactMapPair.getRight().setName(reconstructionStrategy.getNegatedName() + "-" + (i + 1));
                return Stream.of(contactMapPair.getLeft(), contactMapPair.getRight());
            }
        })).filter(reconstructionContactMap -> reconstructionContactMap.getNumberOfContacts() > 0).collect(Collectors.toList());
        Map<String, List<Future<ReconstructionResult>>> reconstructionFutures = new HashMap<>();
        for (ReconstructionContactMap contactMap : contactMaps) {
            String name = contactMap.getName().split("-")[0];
            logger.info("[{}] handling contact map definition {}", explorerChain.getStfId(), name);
            if (!reconstructionFutures.containsKey(name)) {
                reconstructionFutures.put(name, new ArrayList<>());
            }
            List<Future<ReconstructionResult>> bin = reconstructionFutures.get(name);
            bin.add(executorService.submit(new ConfoldServiceWorker("/home/sb/programs/confold_v1.0/confold.pl", contactMap.getSequence(), contactMap.getSecondaryStructureElements(), contactMap.getCaspRRRepresentation(), contactMap.getConfoldRRType())));
        }
        for (Map.Entry<String, List<Future<ReconstructionResult>>> reconstructionFuture : reconstructionFutures.entrySet()) {
            try {
                String name = reconstructionFuture.getKey();
                List<Chain> reconstructions = reconstructionFuture.getValue().stream().map(future -> {
                    try {
                        return future.get();
                    } catch (Exception e) {
                        throw new ComputationException(e);
                    }
                }).map(ReconstructionResult::getChains).flatMap(Collection::stream).collect(Collectors.toList());
                logger.info("[{}][{}] {} reconstructs in bin", explorerChain.getStfId(), name, reconstructions.size());
                List<TMAlignAlignmentResult> alignmentResults = new ArrayList<>();
                List<Path> tmpFiles = new ArrayList<>();
                if (reconstructions.isEmpty()) {
                    throw new ComputationException("reconstruction did not yield any reconstructs");
                }
                for (Chain reconstructedChain : reconstructions) {
                    Path reconstructPath = Files.createTempFile("confoldservice-recon", ".pdb");
                    tmpFiles.add(reconstructPath);
                    Files.write(reconstructPath, reconstructedChain.getPdbRepresentation().getBytes());
                    alignmentResults.add(TM_ALIGN_SERVICE.process(new String[] { "/home/sb/programs/tmalign", nativeChainPath.toFile().getAbsolutePath(), reconstructPath.toFile().getAbsolutePath() }));
                }
                logger.info("[{}][{}] {} alignments in bin", explorerChain.getStfId(), name, alignmentResults.size());
                if (alignmentResults.isEmpty()) {
                    throw new ComputationException("tmalign did not yield any alignments");
                }
                for (TMAlignAlignmentResult alignmentResult : alignmentResults) {
                    double rmsd = alignmentResult.getRootMeanSquareDeviation().getScore();
                    String line = explorerChain.getStfId() + "," + name + "," + rmsd;
                    logger.info("[{}][{}] {}", explorerChain.getStfId(), name, line);
                    fileWriter.write(line + System.lineSeparator());
                    fileWriter.flush();
                }
                // cleanup
                for (Path tmpFile : tmpFiles) {
                    Files.delete(tmpFile);
                }
            } catch (IOException e) {
                throw new ComputationException(e);
            }
        }
    } catch (IOException | AlignmentException e) {
        throw new ComputationException(e);
    }
}

Example 2 with ContactStructuralInformation

use of de.bioforscher.jstructure.efr.model.si.ContactStructuralInformation 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 3 with ContactStructuralInformation

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

the class StructuralInformationParserServiceTest method shouldParseStructuralInformationFile.

@Test
public void shouldParseStructuralInformationFile() {
    Chain chain = StructureParser.fromPdbId("1bdd").parse().getFirstChain();
    List<AminoAcid> aminoAcids = chain.aminoAcids().collect(Collectors.toList());
    Start2FoldXmlParser.parseSpecificExperiment(chain, TestUtils.getResourceAsInputStream("efr/STF0045.xml"), Stream.of(185).collect(Collectors.toList()));
    List<AminoAcid> earlyFoldingResidues = chain.aminoAcids().filter(aminoAcid -> aminoAcid.getFeature(Start2FoldResidueAnnotation.class).isEarly()).collect(Collectors.toList());
    List<ContactStructuralInformation> contactStructuralInformation = StructuralInformationParserService.getInstance().parseContactStructuralInformationFile(TestUtils.getResourceAsInputStream("si/STF0045.out"), chain, earlyFoldingResidues);
    contactStructuralInformation.stream().map(ContactStructuralInformation::getCsvLine).forEach(System.out::println);
    System.out.println();
    List<ResidueStructuralInformation> residueStructuralInformation = StructuralInformationParserService.getInstance().composeResidueStructuralInformation(aminoAcids, earlyFoldingResidues, contactStructuralInformation);
    residueStructuralInformation.stream().map(ResidueStructuralInformation::getCsvLine).forEach(System.out::println);
}

Example 4 with ContactStructuralInformation

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

the class A02_WriteHeatmapFiles method handleChain.

private static void handleChain(ExplorerChain explorerChain) {
    try {
        String entryId = explorerChain.getStfId();
        System.out.println(entryId);
        Chain chain = explorerChain.getChain();
        List<ContactStructuralInformation> contactStructuralInformation = explorerChain.getContacts();
        StringJoiner stringJoiner = new StringJoiner(System.lineSeparator(), "res1,res2,avg" + System.lineSeparator(), "");
        List<AminoAcid> aminoAcids = chain.aminoAcids().collect(Collectors.toList());
        SetOperations.cartesianProductOf(aminoAcids, aminoAcids).forEach(pair -> {
            int residueIdentifier1 = pair.getLeft().getResidueIdentifier().getResidueNumber();
            int residueIdentifier2 = pair.getRight().getResidueIdentifier().getResidueNumber();
            Optional<ContactStructuralInformation> information = contactStructuralInformation.stream().filter(contact -> (contact.getResidueIdentifier1() == residueIdentifier1 && contact.getResidueIdentifier2() == residueIdentifier2) || (contact.getResidueIdentifier1() == residueIdentifier2 && contact.getResidueIdentifier2() == residueIdentifier1)).findFirst();
            if (information.isPresent()) {
                stringJoiner.add(residueIdentifier1 + "," + residueIdentifier2 + "," + information.get().getAverageRmsdIncrease());
            } else {
                stringJoiner.add(residueIdentifier1 + "," + residueIdentifier2 + "," + "0.0");
            }
        });
        Files.write(Start2FoldConstants.DATA_DIRECTORY.resolve("si").resolve("heatmaps").resolve(entryId + ".csv"), stringJoiner.toString().getBytes());
    } catch (Exception e) {
        e.printStackTrace();
    }
}

Example 5 with ContactStructuralInformation

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

the class EvolutionaryCouplingParser method parsePlmScore.

private static void parsePlmScore(ContactStructuralInformation contact, Element table) {
    String residueIdentifier1 = String.valueOf(contact.getResidueIdentifier1());
    String residueIdentifier2 = String.valueOf(contact.getResidueIdentifier2());
    OptionalDouble plmScore = table.getElementsByTag("tr").stream().skip(1).map(element -> element.getElementsByTag("td")).filter(elements -> (elements.get(2).text().equals(residueIdentifier1) && elements.get(4).text().equals(residueIdentifier2)) || (elements.get(2).text().equals(residueIdentifier2) && elements.get(4).text().equals(residueIdentifier1))).mapToDouble(elements -> Double.valueOf(elements.get(6).text())).mapToObj(StandardFormat::format).mapToDouble(Double::valueOf).findFirst();
    OptionalInt rank = table.getElementsByTag("tr").stream().skip(1).map(element -> element.getElementsByTag("td")).filter(elements -> (elements.get(2).text().equals(residueIdentifier1) && elements.get(4).text().equals(residueIdentifier2)) || (elements.get(2).text().equals(residueIdentifier2) && elements.get(4).text().equals(residueIdentifier1))).mapToInt(elements -> Integer.valueOf(elements.get(0).text())).findFirst();
    contact.setPlmScore(plmScore.orElse(0.0));
    contact.setCouplingRank(rank.orElse(-1));
// contact.setCouplingRank(rank.orElse(Integer.MAX_VALUE));
}