Examples with DisequilibriumElement org.dash.valid.DisequilibriumElement used on opensource projects

Example 1 with DisequilibriumElement

use of org.dash.valid.DisequilibriumElement in project ImmunogeneticDataTools by nmdp-bioinformatics.

the class HLAFrequenciesLoader method loadNMDPLinkageReferenceData.

public static List<DisequilibriumElement> loadNMDPLinkageReferenceData(InputStream inStream, Locus[] locusPositions) throws IOException, InvalidFormatException {
    List<DisequilibriumElement> disequilibriumElements = new ArrayList<DisequilibriumElement>();
    Workbook workbook = WorkbookFactory.create(inStream);
    // Return first sheet from the XLSX workbook
    Sheet mySheet = workbook.getSheetAt(0);
    // Get iterator to all the rows in current sheet
    Iterator<Row> rowIterator = mySheet.iterator();
    int firstRow = mySheet.getFirstRowNum();
    List<String> raceHeaders = null;
    // Traversing over each row of XLSX file
    while (rowIterator.hasNext()) {
        Row row = rowIterator.next();
        if (row.getRowNum() == firstRow) {
            raceHeaders = readHeaderElementsByRace(row);
        } else {
            disequilibriumElements.add(readDiseqilibriumElementsByRace(row, raceHeaders, locusPositions));
        }
    }
    workbook.close();
    return disequilibriumElements;
}

Example 2 with DisequilibriumElement

use of org.dash.valid.DisequilibriumElement in project ImmunogeneticDataTools by nmdp-bioinformatics.

the class HLAFrequenciesLoader method loadStandardReferenceData.

public static List<DisequilibriumElement> loadStandardReferenceData(BufferedReader reader) throws IOException {
    String row;
    String[] columns;
    HashMap<String, List<FrequencyByRace>> frequencyMap = new HashMap<String, List<FrequencyByRace>>();
    while ((row = reader.readLine()) != null) {
        columns = row.split(GLStringConstants.COMMA);
        String race = columns[0];
        String haplotype = columns[1];
        Double frequency = new Double(columns[2]);
        String rank = columns[3];
        List<FrequencyByRace> freqList = frequencyMap.get(haplotype);
        if (freqList == null) {
            freqList = new ArrayList<FrequencyByRace>();
        }
        FrequencyByRace freqByRace = new FrequencyByRace(frequency, rank, race);
        freqList.add(freqByRace);
        frequencyMap.put(haplotype, freqList);
    }
    List<DisequilibriumElement> disequilibriumElements = new ArrayList<DisequilibriumElement>();
    DisequilibriumElementByRace disElement;
    HashMap<String, Locus> locusMap = new HashMap<String, Locus>();
    Locus locus = null;
    for (String haplotype : frequencyMap.keySet()) {
        String[] locusHaplotypes = haplotype.split(GLStringConstants.GENE_PHASE_DELIMITER);
        HashMap<Locus, List<String>> hlaElementMap = new HashMap<Locus, List<String>>();
        for (String locusHaplotype : locusHaplotypes) {
            String[] parts = locusHaplotype.split(GLStringUtilities.ESCAPED_ASTERISK);
            List<String> val = new ArrayList<String>();
            val.add(locusHaplotype);
            if (locusMap.containsKey(parts[0])) {
                locus = locusMap.get(parts[0]);
            } else {
                locus = Locus.normalizeLocus(Locus.lookup(parts[0]));
                locusMap.put(parts[0], locus);
            }
            hlaElementMap.put(locus, val);
        }
        disElement = new DisequilibriumElementByRace(hlaElementMap, frequencyMap.get(haplotype));
        disequilibriumElements.add(disElement);
    }
    reader.close();
    return disequilibriumElements;
}

Example 3 with DisequilibriumElement

use of org.dash.valid.DisequilibriumElement in project ImmunogeneticDataTools by nmdp-bioinformatics.

the class HLAFrequenciesLoader method init.

private void init(Set<File> frequencyFiles, File allelesFile) {
    Set<Linkages> linkages = new HashSet<Linkages>();
    try {
        for (File frequencyFile : frequencyFiles) {
            InputStream is = new FileInputStream(frequencyFile);
            InputStreamReader isr = new InputStreamReader(is);
            BufferedReader reader = new BufferedReader(isr);
            List<DisequilibriumElement> elements = loadStandardReferenceData(reader);
            EnumSet<Locus> loci = Locus.lookup(elements.iterator().next().getLoci());
            linkages.addAll(Linkages.lookup(loci));
            this.disequilibriumElementsMap.put(loci, elements);
        }
        LinkagesLoader.getInstance(linkages);
        if (allelesFile != null) {
            loadIndividualLocusFrequencies(allelesFile);
        }
    } catch (IOException e) {
        LOGGER.severe("Couldn't load disequilibrium element reference file.");
        e.printStackTrace();
        System.exit(-1);
    }
}

Example 4 with DisequilibriumElement

use of org.dash.valid.DisequilibriumElement in project ImmunogeneticDataTools by nmdp-bioinformatics.

the class HLAFrequenciesLoader method loadLinkageReferenceData.

public List<DisequilibriumElement> loadLinkageReferenceData(String filename, Locus[] locusPositions) throws FileNotFoundException, IOException {
    BufferedReader reader = new BufferedReader(new InputStreamReader(HLAFrequenciesLoader.class.getClassLoader().getResourceAsStream(filename)));
    String row;
    String[] columns;
    HashMap<Locus, List<String>> hlaElementMap;
    List<DisequilibriumElement> disequilibriumElements = new ArrayList<DisequilibriumElement>();
    while ((row = reader.readLine()) != null) {
        hlaElementMap = new HashMap<Locus, List<String>>();
        columns = row.split(GLStringConstants.TAB);
        for (int i = 0; i < locusPositions.length; i++) {
            List<String> val = new ArrayList<String>();
            val.add(GLStringConstants.DASH.equals(columns[i]) ? GLStringConstants.NNNN : columns[i]);
            hlaElementMap.put(locusPositions[i], val);
        }
        disequilibriumElements.add(new BaseDisequilibriumElement(hlaElementMap, columns[locusPositions.length], columns[locusPositions.length + 1]));
    }
    reader.close();
    return disequilibriumElements;
}

Example 5 with DisequilibriumElement

use of org.dash.valid.DisequilibriumElement in project ImmunogeneticDataTools by nmdp-bioinformatics.

the class NormalizeFrequencyFile method call.

@Override
public Integer call() throws Exception {
    PrintWriter writer = new PrintWriter(outputFile);
    if (SINGLE.equals(frequencies)) {
        List<String> singleLocusFrequencies = HLAFrequenciesLoader.loadIndividualLocusFrequency(new FileInputStream(inputFile));
        for (String allele : singleLocusFrequencies) {
            writer.write(allele + GLStringConstants.NEWLINE);
        }
    } else {
        HashSet<String> linkageNames = new HashSet<String>();
        linkageNames.add(frequencies);
        Set<Linkages> linkagesSet = Linkages.lookup(linkageNames);
        LinkagesLoader.getInstance(linkagesSet);
        List<DisequilibriumElement> disequilibriumElements = HLAFrequenciesLoader.loadNMDPLinkageReferenceData(new FileInputStream(inputFile), LOCUS_POSITION_MAP.get(Linkages.lookup(frequencies).getLoci()));
        for (DisequilibriumElement element : disequilibriumElements) {
            StringBuffer sb = new StringBuffer();
            int locusCounter = 0;
            for (Locus locus : Locus.lookup(element.getLoci())) {
                if (locusCounter > 0) {
                    sb.append(GLStringConstants.GENE_PHASE_DELIMITER);
                }
                sb.append(element.getHlaElement(locus).get(0));
                locusCounter++;
            }
            List<FrequencyByRace> frequencies = ((DisequilibriumElementByRace) element).getFrequenciesByRace();
            for (FrequencyByRace frequency : frequencies) {
                writer.write(frequency.getRace() + GLStringConstants.COMMA + sb + GLStringConstants.COMMA + frequency.getFrequency() + GLStringConstants.COMMA + frequency.getRank() + GLStringConstants.NEWLINE);
            }
        }
    }
    writer.close();
    return 0;
}