public static void convertGMTFileToPlainTextBinaryNetwork(
String infile, String outfile, boolean hasIds, boolean isActuallyGMT) throws IOException {
TextFile in = new TextFile(infile, TextFile.R);
String line = null;
Map<String, Integer> hashItems = new HashMap<String, Integer>();
Map<String, Set<Integer>> hashSetIndices = new HashMap<String, Set<Integer>>();
List<String> sets = new ArrayList<String>();
int nextIndex = 0;
int nrItemSets = 0;
while ((line = in.readLine()) != null) {
String[] split = line.split("\t");
if (split.length == 0) {
continue;
}
int firstItemIndex = hasIds ? (isActuallyGMT ? 2 : 1) : 0;
if (hasIds) {
sets.add(split[0]);
}
for (int i = firstItemIndex; i < split.length; i++) {
Integer itemIndex = hashItems.get(split[i]);
if (itemIndex == null) {
hashItems.put(split[i], nextIndex);
nextIndex++;
}
Set<Integer> setIndicesThisItem = hashSetIndices.get(split[i]);
if (setIndicesThisItem == null) {
setIndicesThisItem = new HashSet<Integer>();
hashSetIndices.put(split[i], setIndicesThisItem);
}
setIndicesThisItem.add(nrItemSets);
}
nrItemSets++;
}
in.close();
TextFile out = new TextFile(outfile, TextFile.W);
for (int i = 0; i < nrItemSets; i++) {
if (hasIds) {
out.write("\t" + sets.get(i));
} else {
out.write("\tComplex" + (i + 1));
}
}
out.writeln();
for (String item : hashSetIndices.keySet()) {
if (!"".equals(item)) {
out.write(item);
Set<Integer> setIndicesThisItem = hashSetIndices.get(item);
for (int set = 0; set < nrItemSets; set++) {
if (setIndicesThisItem.contains(set)) {
out.write("\t1");
} else {
out.write("\t0");
}
}
out.writeln();
}
}
out.close();
}
private static void remapPositionsAndSnps(
LinkedHashSet<String> OrderingEst,
HashMap<String, String> mappingEst,
HashMap<String, String> mappingUmcg,
String outFile,
String outFile2) {
try {
TextFile out = new TextFile(outFile, TextFile.W);
TextFile out2 = new TextFile(outFile2, TextFile.W);
for (String entry : OrderingEst) {
// System.out.println(entry);
String snpPos = mappingEst.get(entry);
String[] t = SPLIT_ON_COLON.split(snpPos);
if (mappingUmcg.containsKey(snpPos)) {
out.writeln(mappingUmcg.get(snpPos));
out2.writeln(t[0] + "\t" + t[1] + "\t" + mappingUmcg.get(snpPos));
} else {
System.out.println("Problem: " + entry);
out.writeln(entry);
out2.writeln(t[0] + "\t" + t[1] + "\t" + entry);
}
}
out.close();
out2.close();
} catch (IOException ex) {
Logger.getLogger(ConverteMappingAndSnpFile.class.getName()).log(Level.SEVERE, null, ex);
}
}
public static void convertPlainTextDoubleMatrixToPlainTextEdgeList(
String matrixfile, String outfile, String infileDelimiter, String outfileDelimiter)
throws IOException {
TextFile in = new TextFile(matrixfile, TextFile.R);
String line = in.readLine();
String[] split = line.split(infileDelimiter);
String[] nodes = Arrays.copyOfRange(split, 1, split.length);
LOGGER.log(Level.FINE, "{0} nodes in {1}", new Object[] {nodes.length, matrixfile});
TextFile out = new TextFile(outfile, TextFile.W);
int lineNr = 1;
int nrEdges = 0;
while ((line = in.readLine()) != null) {
split = line.split(infileDelimiter);
if (split.length != nodes.length + 1) {
throw new IllegalArgumentException(
"The data in file '"
+ matrixfile
+ "' are not a matrix. Check line "
+ (lineNr + 1)
+ ".");
}
String node = split[0];
for (int i = 1; i < split.length; i++) {
if (lineNr == i) {
continue; // skip diagonal
}
try {
double weight = Double.parseDouble(split[i]);
if (weight > 0) {
out.writeln(node + outfileDelimiter + nodes[i - 1] + outfileDelimiter + weight);
nrEdges++;
}
} catch (NumberFormatException ex) {
throw new IllegalArgumentException(
"The data in file '"
+ matrixfile
+ "' are not numerical (cast to double failed). Check line "
+ (lineNr + 1)
+ ", column "
+ (i + 1)
+ ".");
}
}
lineNr++;
}
in.close();
out.close();
LOGGER.log(
Level.FINE,
"{0} edges from {1} written to ''{2}''",
new Object[] {nrEdges, matrixfile, outfile});
}
public static void writeGMTFileBasedOnGeneSetFileAndMappingFileRemovingDuplicateGeneSets(
String genesetfile, String mappingfile, String mappingdelimiter, String gmtfile)
throws IOException {
TextFile in = new TextFile(mappingfile, TextFile.R);
String line = in.readLine();
Map<String, String> code2name = new HashMap<String, String>();
while ((line = in.readLine()) != null) {
String[] split = line.split(mappingdelimiter);
String name = split[2].trim().replace("\"", "") + " (" + split[6].trim() + ")";
code2name.put(split[0].trim().replace("\"", ""), name);
}
in.close();
System.out.println(code2name.size() + " gene set annotations read");
in = new TextFile(genesetfile, TextFile.R);
TextFile out = new TextFile(gmtfile, TextFile.W);
Map<String, Integer> usedNames = new HashMap<String, Integer>();
Set<String> usedGeneSets = new HashSet<String>();
while ((line = in.readLine()) != null) {
String[] split = line.split("\t");
if (split.length == 0) {
continue;
}
String code = split[0].trim();
String name = code2name.get(code);
if (name == null) {
LOGGER.log(Level.WARNING, "No annotation for gene set ''{0}''", code);
} else {
String genes = Arrays.asList(Arrays.copyOfRange(split, 1, split.length)).toString();
System.out.println(genes);
if (!usedGeneSets.contains(genes)) {
Integer oldNrItems = usedNames.get(name);
if (oldNrItems == null) {
out.write(code + "\t" + name);
for (int i = 1; i < split.length; i++) {
out.write("\t" + split[i]);
}
out.writeln();
usedNames.put(name, split.length - 1);
usedGeneSets.add(genes);
} else {
// if (oldNrItems != split.length - 1) {
System.out.println("Sets with different numbers of genes for " + name);
// }
}
}
}
}
in.close();
out.close();
}
private static ArrayList<InterChrContact> writeRawInterContactInformation(
ArrayList<InterChrContact> contacts, String fileToWrite) throws IOException {
ArrayList<InterChrContact> chrContactInfo = new ArrayList<InterChrContact>();
TextFile outWriter = new TextFile(fileToWrite, TextFile.W);
String row;
for (InterChrContact contact : contacts) {
outWriter.writeln(
contact.getChrLocationSmaller()
+ "\t"
+ contact.getChrLocationLarger()
+ "\t"
+ contact.getContactValue());
}
outWriter.close();
return chrContactInfo;
}
@Override
public void run() throws IOException {
for (int d = 0; d < m_gg.length; d++) {
// now test all triples
SNPLoader snpLoader = m_gg[d].getGenotypeData().createSNPLoader();
int[] indWGA = m_gg[d].getExpressionToGenotypeIdArray();
for (int perm = 0; perm < m_settings.nrPermutationsFDR + 1; perm++) {
String outfile = null;
if (perm == 0) {
outfile = outDir + m_gg[d].getSettings().name + "_IVAnalysis-RealData.txt";
} else {
outfile =
outDir + m_gg[d].getSettings().name + "_IVAnalysis-PermutationRound-" + perm + ".txt";
m_gg[d].permuteSampleLables();
}
TextFile out = new TextFile(outfile, TextFile.W);
Iterator<Triple<String, String, String>> it = snpProbeCombos.iterator();
Triple<String, String, String> next = it.next();
ProgressBar pb =
new ProgressBar(
snpProbeCombos.size(), "Running Mediation Analysis - Permutation " + perm);
out.writeln(
"SNP\tSNP Chr\tSNP ChrPos\t"
+ "Alleles\tDirectionAllele\t"
+ "N\t"
+ "CisArrayAddress\tCisProbe Chr\tCisProbe ChrPos\t"
+ "CisGeneName\t"
+ "TransArrayAddress\tTransProbe Chr\tTransProbe ChrPos\t"
+ "TransGeneName\t"
+ "CisTrans-Correlation\t"
+ "Cis-eQTL-Beta\t"
+ "Cis-eQTL-SE\t"
+ "CisTrans-Beta\t"
+ "CisTrans-SE\t"
+ "Trans-eQTL-Beta\t"
+ "Trans-eQTL-SE\t"
+ "CisTrans-Residual-Correlation\t"
+ "CisTrans-Residual-Beta\t"
+ "CisTrans-Residual-SE\t"
+ "Trans-eQTL-Residual-Beta\t"
+ "Trans-eQTL-Residual-SE\t"
+ "Beta-Ratio");
while (next != null) {
String snp = next.getLeft();
String cisprobe = next.getMiddle();
String transprobe = next.getRight();
Integer snpId = m_gg[d].getGenotypeData().getSnpToSNPId().get(snp);
Integer cisProbeId = m_gg[d].getExpressionData().getProbeToId().get(cisprobe);
Integer transProbeId = m_gg[d].getExpressionData().getProbeToId().get(transprobe);
if (snpId == -9 || cisProbeId == null || transProbeId == null) {
// out.writeln(snp + "\t" + snpId + "\t" + cisprobe + "\t" +
// cisProbeId + "\t" + null + "\t" + transprobe + "\t" + transProbeId + "\t" + null +
// "\t" + null + "\t" + null + "\t" + null + "\t" + null +
// "\tNA\tNA\tNA\tNA\tNA\tNA\tNA\tNA\tNA\tNA");
} else {
SNP snpObj = m_gg[d].getGenotypeData().getSNPObject(snpId);
snpLoader.loadGenotypes(snpObj);
if (snpLoader.hasDosageInformation()) {
snpLoader.loadDosage(snpObj);
}
double[] origCisVals = m_gg[d].getExpressionData().getMatrix()[cisProbeId];
double[] origTransVals = m_gg[d].getExpressionData().getMatrix()[transProbeId];
int calledGenotypes = 0;
for (int i = 0; i < m_gg[d].getExpressionData().getIndividuals().length; i++) {
int genotypeId = indWGA[i];
short gt = snpObj.getGenotypes()[genotypeId];
if (genotypeId > -1 && gt > -1) {
calledGenotypes++;
}
}
double[] genotypes = new double[calledGenotypes];
double[] cisvals = new double[calledGenotypes];
double[] transvals = new double[calledGenotypes];
calledGenotypes = 0;
for (int i = 0; i < m_gg[d].getExpressionData().getIndividuals().length; i++) {
int genotypeId = indWGA[i];
short gt = snpObj.getGenotypes()[genotypeId];
if (genotypeId > -1 && gt > -1) {
genotypes[calledGenotypes] = snpObj.getDosageValues()[genotypeId];
cisvals[calledGenotypes] = origCisVals[i];
transvals[calledGenotypes] = origTransVals[i];
calledGenotypes++;
}
}
// normalize genotype and cis + trans to get beta's equal to the correlation coefficient
genotypes = normalize(genotypes);
cisvals = normalize(cisvals);
transvals = normalize(transvals);
double corrCisTrans =
JSci.maths.ArrayMath.correlation(cisvals, transvals); // for code validation
double[] cisTransRCs =
Regression.getLinearRegressionCoefficients(
cisvals, transvals); // returns beta, alpha, se, t
double[] snpCisRCs =
Regression.getLinearRegressionCoefficients(
genotypes, cisvals); // returns beta, alpha, se, t
double[] snpTransRCs = Regression.getLinearRegressionCoefficients(genotypes, transvals);
// remove correlation between cis and trans probe
// double[] resCis = new double[cisvals.length];
double[] resTransVals = new double[cisvals.length];
for (int i = 0; i < resTransVals.length; i++) {
// resCis[i] = cisvals[i] - snpCisRCs[0] * genotypes[i];
resTransVals[i] = transvals[i] - cisTransRCs[0] * cisvals[i];
}
resTransVals = normalize(resTransVals);
double[] cisResTransRCs =
Regression.getLinearRegressionCoefficients(
cisvals, resTransVals); // returns beta, alpha, se, t
double[] snpResTransRCs =
Regression.getLinearRegressionCoefficients(genotypes, resTransVals);
double rescorr =
JSci.maths.ArrayMath.correlation(cisvals, resTransVals); // for code validation
out.writeln(
snp
+ "\t"
+ snpObj.getChr()
+ "\t"
+ snpObj.getChrPos()
+ "\t"
+ BaseAnnot.toString(snpObj.getAlleles()[0])
+ "/"
+ BaseAnnot.toString(snpObj.getAlleles()[1])
+ "\t"
+ BaseAnnot.toString(snpObj.getAlleles()[0])
+ "\t"
+ transvals.length
+ "\t"
+ cisprobe
+ "\t"
+ m_gg[d].getExpressionData().getChr()[cisProbeId]
+ "\t"
+ m_gg[d].getExpressionData().getChrStart()[cisProbeId]
+ ":"
+ m_gg[d].getExpressionData().getChrStop()[cisProbeId]
+ "\t"
+ m_gg[d].getExpressionData().getAnnotation()[cisProbeId]
+ "\t"
+ transprobe
+ "\t"
+ m_gg[d].getExpressionData().getChr()[transProbeId]
+ "\t"
+ m_gg[d].getExpressionData().getChrStart()[transProbeId]
+ ":"
+ m_gg[d].getExpressionData().getChrStop()[transProbeId]
+ "\t"
+ m_gg[d].getExpressionData().getAnnotation()[transProbeId]
+ "\t"
+ corrCisTrans
+ "\t"
+ snpCisRCs[0]
+ "\t"
+ snpCisRCs[2]
+ "\t"
+ cisTransRCs[0]
+ "\t"
+ cisTransRCs[2]
+ "\t"
+ snpTransRCs[0]
+ "\t"
+ snpTransRCs[2]
+ "\t"
+ rescorr
+ "\t"
+ cisResTransRCs[0]
+ "\t"
+ cisResTransRCs[2]
+ "\t"
+ snpResTransRCs[0]
+ "\t"
+ snpResTransRCs[2]
+ "\t"
+ (snpResTransRCs[0] / snpTransRCs[0]));
snpObj.clearGenotypes();
}
if (it.hasNext()) {
next = it.next();
} else {
next = null;
}
pb.iterate();
}
pb.close();
out.close();
}
snpLoader.close();
}
}
public final void compareOverlapAndZScoreDirectionTwoEQTLFiles(
String eQTL,
String meQTL,
String eQTMFile,
String outputFile,
boolean matchOnGeneName,
double fdrCutt,
boolean matchSnpOnPos,
boolean splitGeneNames,
boolean flipUsingEQTM,
boolean topeffect)
throws IOException, Exception {
System.out.println("Performing comparison of eQTLs and meQTLs");
double filterOnFDR =
fdrCutt; // Do we want to use another FDR measure? When set to -1 this is not used at all.
HashSet<String> hashExcludeEQTLs =
new HashSet<
String>(); // We can exclude some eQTLs from the analysis. If requested, put the entire
// eQTL string in this HashMap for each eQTL. Does not work in combination
// with mathcing based on chr and pos
HashSet<String> hashConfineAnalysisToSubsetOfProbes =
new HashSet<
String>(); // We can confine the analysis to only a subset of probes. If requested put
// the probe name in this HapMap
HashSet<String> hashTestedSNPsThatPassedQC =
null; // We can confine the analysis to only those eQTLs for which the SNP has been
// successfully passed QC, otherwise sometimes unfair comparisons are made. If
// requested, put the SNP name in this HashMap
// Load the eQTM File
QTLTextFile eQTLsTextFile = new QTLTextFile(eQTMFile, QTLTextFile.R);
HashMap<String, ArrayList<EQTL>> eQtmInfo = new HashMap<String, ArrayList<EQTL>>();
for (Iterator<EQTL> eQtlIt = eQTLsTextFile.getEQtlIterator(); eQtlIt.hasNext(); ) {
EQTL eQtm = eQtlIt.next();
String eQtmKey = eQtm.getRsName();
if (!eQtm.getAlleleAssessed().equals("C")) {
eQtm.setAlleleAssessed("C");
eQtm.setZscore(eQtm.getZscore() * -1);
Double[] zscores = eQtm.getDatasetZScores();
Double[] correlation = eQtm.getCorrelations();
for (int i = 0; i < eQtm.getDatasets().length; ++i) {
zscores[i] *= -1;
correlation[i] *= -1;
}
eQtm.setDatasetZScores(zscores);
eQtm.setCorrelations(correlation);
}
ArrayList<EQTL> posEqtls = eQtmInfo.get(eQtmKey);
if (posEqtls == null) {
posEqtls = new ArrayList<EQTL>(1);
posEqtls.add(eQtm);
eQtmInfo.put(eQtmKey, posEqtls);
} else if (!topeffect) {
eQtmInfo.get(eQtmKey).add(eQtm);
}
}
System.out.println("eQTMs read in: " + eQtmInfo.size());
// Now load the eQTLs for file 1:
THashMap<String, String[]> hashEQTLs = new THashMap<String, String[]>();
THashSet<String> hashUniqueProbes = new THashSet<String>();
THashSet<String> hashUniqueGenes = new THashSet<String>();
TextFile in = new TextFile(eQTL, TextFile.R);
in.readLine();
String[] data = in.readLineElemsReturnReference(SPLIT_ON_TAB);
if (data.length < 5) {
throw new IllegalStateException(
"QTL File does not have enough columns. Detected columns: "
+ data.length
+ " in file "
+ in.getFileName());
}
while (data != null) {
if (filterOnFDR == -1 || Double.parseDouble(data[18]) <= filterOnFDR) {
if (hashConfineAnalysisToSubsetOfProbes.isEmpty()
|| hashConfineAnalysisToSubsetOfProbes.contains(data[4])) {
if (matchOnGeneName) {
if (data[16].length() > 1) {
if (splitGeneNames) {
for (String gene : SEMI_COLON_PATTERN.split(data[16])) {
hashEQTLs.put(
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + gene, data);
hashUniqueProbes.add(data[4]);
hashUniqueGenes.add(gene);
}
} else {
if (!hashExcludeEQTLs.contains(data[1] + "\t" + data[16])) {
hashEQTLs.put(
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + data[16], data);
hashUniqueProbes.add(data[4]);
hashUniqueGenes.add(data[16]);
// log.write("Added eQTL from original file " + (matchSnpOnPos ? data[2] + ":" +
// data[3] : data[1]) + "\t" + data[16]);
}
}
}
} else {
if (!hashExcludeEQTLs.contains(data[1] + "\t" + data[4])) {
hashEQTLs.put(
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + data[4], data);
hashUniqueProbes.add(data[4]);
hashUniqueGenes.add(data[16]);
// log.write("Added eQTL from original file " + (matchSnpOnPos ? data[2] + ":" +
// data[3] : data[1]) + "\t" + data[4]);
}
}
}
data = in.readLineElemsReturnReference(SPLIT_ON_TAB);
}
}
in.close();
int nrUniqueProbes = hashUniqueProbes.size();
int nrUniqueGenes = hashUniqueGenes.size();
hashUniqueProbes = null;
hashUniqueGenes = null;
// Initialize Graphics2D for the Z-Score allelic direction comparison:
// int width = 1000;
// int height = 1000;
// int margin = 100;
// int x0 = margin;
// int x1 = width - margin;
// int y0 = margin;
// int y1 = height - margin;
ZScorePlot zs = new ZScorePlot();
String zsOutFileName = outputFile + "-ZScoreComparison.pdf";
zs.init(2, new String[] {"eQTLs", "meQTLs"}, true, zsOutFileName);
// Variables holding variousStatistics:
int nreQTLsIdenticalDirection = 0;
int nreQTLsOppositeDirection = 0;
HashMap<String, Integer> hashEQTLNrTimesAssessed = new HashMap<String, Integer>();
THashSet<String> hashEQTLs2 = new THashSet<String>();
THashSet<String> hashUniqueProbes2 = new THashSet<String>();
THashSet<String> hashUniqueGenes2 = new THashSet<String>();
THashSet<String> hashUniqueProbesOverlap = new THashSet<String>();
THashSet<String> hashUniqueGenesOverlap = new THashSet<String>();
int counterFile2 = 0;
int overlap = 0;
ArrayDoubleList vecX = new ArrayDoubleList();
ArrayDoubleList vecY = new ArrayDoubleList();
// Vector holding all opposite allelic effects:
// LinkedHashSet<String> vecOppositeEQTLs = new LinkedHashSet<String>();
// Now process file 2:
in = new TextFile(meQTL, TextFile.R);
in.readLine();
int skippedDueToMapping = 0;
data = null;
TextFile identicalOut =
new TextFile(outputFile + "-eQTLsWithIdenticalDirecton.txt.gz", TextFile.W);
TextFile disconcordantOut = new TextFile(outputFile + "-OppositeEQTLs.txt", TextFile.W);
TextFile log = new TextFile(outputFile + "-eQTL-meQTL-ComparisonLog.txt", TextFile.W);
TextFile log2 = new TextFile(outputFile + "-eQTM-missingnessLog.txt", TextFile.W);
THashSet<String> identifiersUsed = new THashSet<String>();
while ((data = in.readLineElemsReturnReference(SPLIT_ON_TAB)) != null) {
if (filterOnFDR == -1 || Double.parseDouble(data[18]) <= filterOnFDR) {
if (!eQtmInfo.containsKey(data[4])) {
skippedDueToMapping++;
log2.write(
"meQTL probe not present In eQTM file:\t"
+ data[4]
+ ", effect statistics: \t"
+ data[0]
+ "\t"
+ data[2]
+ "\t"
+ data[3]
+ "\t"
+ data[16]
+ "\n");
continue;
}
String orgDataFour = data[4];
for (int i = 0; i < eQtmInfo.get(orgDataFour).size(); ++i) {
if (topeffect && i > 0) {
break;
}
data[16] = eQtmInfo.get(orgDataFour).get(i).getProbeHUGO();
data[4] = eQtmInfo.get(orgDataFour).get(i).getProbe();
if (flipUsingEQTM) {
Double zScoreQTM = eQtmInfo.get(orgDataFour).get(i).getZscore();
if (zScoreQTM < 0) {
data[10] = String.valueOf(Double.parseDouble(data[10]) * -1);
}
}
if (hashConfineAnalysisToSubsetOfProbes.isEmpty()
|| hashConfineAnalysisToSubsetOfProbes.contains(data[4])) {
if (matchOnGeneName) {
if (!hashExcludeEQTLs.contains(data[1] + "\t" + data[16])) {
if (data[16].length() > 1) {
if (splitGeneNames) {
for (String gene : SEMI_COLON_PATTERN.split(data[16])) {
hashUniqueProbes2.add(data[4]);
hashUniqueGenes2.add(gene);
if (!hashEQTLs2.contains(
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + gene)) {
hashEQTLs2.add(
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + gene);
counterFile2++;
}
}
} else {
hashUniqueProbes2.add(data[4]);
hashUniqueGenes2.add(data[16]);
if (!hashEQTLs2.contains(
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + data[16])) {
hashEQTLs2.add(
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + data[16]);
counterFile2++;
}
}
}
}
} else {
if (!hashExcludeEQTLs.contains(data[1] + "\t" + data[4])) {
// hashEQTLs2.put(data[1] + "\t" + data[4], str);
hashUniqueProbes2.add(data[4]);
hashUniqueGenes2.add(data[16]);
counterFile2++;
}
}
}
String[] QTL = null;
String identifier = null;
if (matchOnGeneName) {
if (data.length > 16 && data[16].length() > 1) {
if (splitGeneNames) {
// NB Plotting and processing of all QTLs here is not okay!
for (String gene : SEMI_COLON_PATTERN.split(data[16])) {
if (!hashExcludeEQTLs.contains(data[1] + "\t" + gene)) {
identifier = (matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + gene;
if (hashEQTLs.containsKey(identifier)) {
QTL = hashEQTLs.get(identifier);
}
}
}
} else {
if (!hashExcludeEQTLs.contains(data[1] + "\t" + data[16])) {
identifier =
(matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + data[16];
if (hashEQTLs.containsKey(identifier)) {
QTL = hashEQTLs.get(identifier);
}
}
}
}
} else {
if (!hashExcludeEQTLs.contains(data[1] + "\t" + data[4])) {
identifier = (matchSnpOnPos ? data[2] + ":" + data[3] : data[1]) + "\t" + data[4];
if (hashEQTLs.containsKey(identifier)) {
QTL = hashEQTLs.get(identifier);
}
}
}
if (QTL == null) {
// The eQTL, present in file 2 is not present in file 1:
// if (Double.parseDouble(data[0]); < 1E-4) {
if (hashTestedSNPsThatPassedQC == null
|| hashTestedSNPsThatPassedQC.contains(data[1])) {
log.write(
"eQTL Present In New file But Not In Original File:\t"
+ identifier
+ "\t"
+ data[0]
+ "\t"
+ data[2]
+ "\t"
+ data[3]
+ "\t"
+ data[16]
+ "\n");
}
// }
double zScore2 = Double.parseDouble(data[10]);
// int posX = 500 + (int) 0;
// int posY = 500 - (int) Math.round(zScore2 * 10);
zs.draw(null, zScore2, 0, 1);
} else {
identifiersUsed.add(identifier);
String[] eQtlData = QTL;
boolean identicalProbe = true;
String probe = data[4];
String probeFound = eQtlData[4];
if (!probe.equals(probeFound)) {
identicalProbe = false;
}
hashUniqueProbesOverlap.add(data[4]);
hashUniqueGenesOverlap.add(data[16]);
if (!hashEQTLNrTimesAssessed.containsKey(identifier)) {
hashEQTLNrTimesAssessed.put(identifier, 1);
} else {
hashEQTLNrTimesAssessed.put(identifier, 1 + hashEQTLNrTimesAssessed.get(identifier));
}
String alleles = eQtlData[8];
String alleleAssessed = eQtlData[9];
String correlations[] = (eQtlData[17]).split(";");
double correlation = 0;
int numCorr1 = 0;
for (int c = 0; c < correlations.length; c++) {
try {
if (!correlations[c].equals("-")) {
correlation += Double.parseDouble(correlations[c]);
numCorr1++;
}
} catch (Exception e) {
}
}
correlation /= (double) numCorr1;
// if(numCorr1 == 0){
// System.out.println("Warning: no correlations defined for
// eqtl file 1");
// }
double zScore = Double.parseDouble(eQtlData[10]);
// double pValue = Double.parseDouble(eQtlData[0]);
String alleles2 = data[8];
String alleleAssessed2 = data[9];
double zScore2 = Double.parseDouble(data[10]);
// double pValue2 = Double.parseDouble(data[0]);
String correlations2[] = data[17].split(";");
double correlation2 = 0;
boolean alleleflipped = false;
if (!alleleAssessed.equals(data[9])) {
if (data[9].equals(eQtlData[8].split("/")[0])) {
alleleflipped = true;
} else {
// System.out.println("WTF BBQ!");
}
}
int numCorr2 = 0;
for (int c = 0; c < correlations2.length; c++) {
try {
if (!correlations2[c].equals("-")) {
correlation2 += (Double.parseDouble(correlations2[c]));
numCorr2++;
}
} catch (NumberFormatException e) {
}
}
// if(numCorr2 == 0){
// System.out.println("Warning: no correlations defined for
// eqtl file 2");
// }
correlation2 /= (double) numCorr2;
if (alleleflipped) {
correlation2 = -correlation2;
}
boolean sameDirection = false;
int nrIdenticalAlleles = 0;
if (alleles.length() > 2 && alleles2.length() > 2) {
for (int a = 0; a < 3; a++) {
for (int b = 0; b < 3; b++) {
if (a != 1 && b != 1) {
if (alleles.getBytes()[a] == alleles2.getBytes()[b]) {
nrIdenticalAlleles++;
}
}
}
}
}
if (nrIdenticalAlleles == 0) {
alleles2 =
(char) BaseAnnot.getComplement((byte) alleles2.charAt(0))
+ "/"
+ (char) BaseAnnot.getComplement((byte) alleles2.charAt(2));
alleleAssessed2 = BaseAnnot.getComplement(alleleAssessed2);
if (alleles.length() > 2 && alleles2.length() > 2) {
for (int a = 0; a < 3; a++) {
for (int b = 0; b < 3; b++) {
if (a != 1 && b != 1) {
if (alleles.getBytes()[a] == alleles2.getBytes()[b]) {
nrIdenticalAlleles++;
}
}
}
}
}
}
if (nrIdenticalAlleles != 2) {
log.write(
"Error! SNPs have incompatible alleles!!:\t"
+ alleles
+ "\t"
+ alleles2
+ "\t"
+ identifier
+ "\n");
} else {
overlap++;
if (!alleleAssessed.equals(alleleAssessed2)) {
zScore2 = -zScore2;
// correlation2 = -correlation2;
alleleAssessed2 = alleleAssessed;
}
// Recode alleles:
// if contains T, but no A, take complement
// if (alleles.contains("T") && !alleles.contains("A")) {
// alleles = BaseAnnot.getComplement(alleles);
// alleleAssessed =
// BaseAnnot.getComplement(alleleAssessed);
// alleleAssessed2 =
// BaseAnnot.getComplement(alleleAssessed2);
// }
if (zScore2 * zScore > 0) {
sameDirection = true;
}
// if(correlation != correlation2 && (numCorr1 > 0 && numCorr2 >
// 0)){
// if(Math.abs(correlation - correlation2) > 0.00001){
// System.out.println("Correlations are different:
// "+lineno+"\t"+correlation +"\t"+correlation2+"\t"+str);
// }
//
// }
zs.draw(zScore, zScore2, 0, 1);
if (!sameDirection) {
nreQTLsOppositeDirection++;
if (matchOnGeneName) {
disconcordantOut.append(
data[1]
+ '\t'
+ data[16]
+ '\t'
+ alleles
+ '\t'
+ alleleAssessed
+ '\t'
+ zScore
+ '\t'
+ alleles2
+ '\t'
+ alleleAssessed2
+ '\t'
+ zScore2);
} else {
disconcordantOut.append(
data[1]
+ '\t'
+ data[4]
+ '\t'
+ alleles
+ '\t'
+ alleleAssessed
+ '\t'
+ zScore
+ '\t'
+ alleles2
+ '\t'
+ alleleAssessed2
+ '\t'
+ zScore2);
}
// int posX = 500 + (int) Math.round(zScore * 10);
// int posY = 500 - (int) Math.round(zScore2 * 10);
vecX.add(zScore);
vecY.add(zScore2);
} else {
// write to output
identicalOut.writeln(
identifier
+ '\t'
+ alleles
+ '\t'
+ alleleAssessed
+ '\t'
+ zScore
+ '\t'
+ alleles2
+ '\t'
+ alleleAssessed2
+ '\t'
+ zScore2);
nreQTLsIdenticalDirection++;
if (alleles.length() > 2
&& !alleles.equals("A/T")
&& !alleles.equals("T/A")
&& !alleles.equals("C/G")
&& !alleles.equals("G/C")) {
// int posX = 500 + (int) Math.round(zScore * 10);
// int posY = 500 - (int) Math.round(zScore2 * 10);
vecX.add(zScore);
vecY.add(zScore2);
}
}
}
}
}
}
}
identicalOut.close();
disconcordantOut.close();
in.close();
log2.close();
log.write(
"\n/// Writing missing QTLs observed in original file but not in the new file ////\n\n");
for (Entry<String, String[]> QTL : hashEQTLs.entrySet()) {
if (!identifiersUsed.contains(QTL.getKey())) {
// The eQTL, present in file 1 is not present in file 2:
// if (Double.parseDouble(QTL.getValue()[0]) < 1E-4) {
if (hashTestedSNPsThatPassedQC == null || hashTestedSNPsThatPassedQC.contains(data[1])) {
log.write(
"eQTL Present In Original file But Not In New File:\t"
+ QTL.getKey()
+ "\t"
+ QTL.getValue()[0]
+ "\t"
+ QTL.getValue()[2]
+ "\t"
+ QTL.getValue()[3]
+ "\t"
+ QTL.getValue()[16]
+ "\n");
}
// }
double zScore = Double.parseDouble(QTL.getValue()[10]);
// int posX = 500 + (int) 0;
// int posY = 500 - (int) Math.round(zScore * 10);
zs.draw(zScore, null, 0, 1);
}
}
log.close();
zs.write(zsOutFileName);
double[] valsX = vecX.toArray();
double[] valsY = vecY.toArray();
if (valsX.length > 2) {
double correlation = JSci.maths.ArrayMath.correlation(valsX, valsY);
double r2 = correlation * correlation;
cern.jet.random.tdouble.engine.DoubleRandomEngine randomEngine =
new cern.jet.random.tdouble.engine.DRand();
cern.jet.random.tdouble.StudentT tDistColt =
new cern.jet.random.tdouble.StudentT(valsX.length - 2, randomEngine);
double pValuePearson = 1;
double tValue = correlation / (Math.sqrt((1 - r2) / (double) (valsX.length - 2)));
if (tValue < 0) {
pValuePearson = tDistColt.cdf(tValue);
} else {
pValuePearson = tDistColt.cdf(-tValue);
}
pValuePearson *= 2;
System.out.println(
"\nCorrelation between the Z-Scores of the overlapping set of eQTLs:\t"
+ correlation
+ "\tP-Value:\t"
+ pValuePearson);
}
TextFile outSummary = new TextFile(outputFile + "-Summary.txt", TextFile.W);
System.out.println("");
System.out.println(
"Nr of eQTLs:\t"
+ hashEQTLs.size()
+ "\tin file:\t"
+ eQTL
+ "\tNrUniqueProbes:\t"
+ nrUniqueProbes
+ "\tNrUniqueGenes:\t"
+ nrUniqueGenes);
outSummary.writeln(
"Nr of eQTLs:\t"
+ hashEQTLs.size()
+ "\tin file:\t"
+ eQTL
+ "\tNrUniqueProbes:\t"
+ nrUniqueProbes
+ "\tNrUniqueGenes:\t"
+ nrUniqueGenes);
System.out.println(
"Nr of meQTLs:\t"
+ counterFile2
+ "\tin file:\t"
+ meQTL
+ "\tNrUniqueProbes:\t"
+ hashUniqueProbes2.size()
+ "\tNrUniqueGenes:\t"
+ hashUniqueGenes2.size()
+ " *With eQTM mapping.");
outSummary.writeln(
"Nr of meQTLs:\t"
+ counterFile2
+ "\tin file:\t"
+ meQTL
+ "\tNrUniqueProbes:\t"
+ hashUniqueProbes2.size()
+ "\tNrUniqueGenes:\t"
+ hashUniqueGenes2.size()
+ " *With eQTM mapping.");
System.out.println("Skipped over meQTLs:\t" + skippedDueToMapping);
outSummary.writeln("Skipped over meQTLs:\t" + skippedDueToMapping);
System.out.println(
"Overlap:\t"
+ overlap
+ "\tNrUniqueProbesOverlap:\t"
+ hashUniqueProbesOverlap.size()
+ "\tNrUniqueGenesOverlap:\t"
+ hashUniqueGenesOverlap.size());
outSummary.writeln(
"Overlap:\t"
+ overlap
+ "\tNrUniqueProbesOverlap:\t"
+ hashUniqueProbesOverlap.size()
+ "\tNrUniqueGenesOverlap:\t"
+ hashUniqueGenesOverlap.size());
System.out.println("");
outSummary.writeln();
System.out.println("Nr eQTLs with identical direction:\t" + nreQTLsIdenticalDirection);
outSummary.writeln("Nr eQTLs with identical direction:\t" + nreQTLsIdenticalDirection);
double proportionOppositeDirection =
100d
* (double) nreQTLsOppositeDirection
/ (double) (nreQTLsOppositeDirection + nreQTLsIdenticalDirection);
String proportionOppositeDirectionString =
(new java.text.DecimalFormat(
"0.00;-0.00", new java.text.DecimalFormatSymbols(java.util.Locale.US)))
.format(proportionOppositeDirection);
System.out.println(
"Nr eQTLs with opposite direction:\t"
+ nreQTLsOppositeDirection
+ "\t("
+ proportionOppositeDirectionString
+ "%)");
outSummary.writeln(
"Nr eQTLs with opposite direction:\t"
+ nreQTLsOppositeDirection
+ "\t("
+ proportionOppositeDirectionString
+ "%)");
outSummary.close();
nrShared = hashUniqueProbesOverlap.size();
nrOpposite = nreQTLsOppositeDirection;
}
private static void processNormalizedIntraContactInformation(
String fileToRead,
String baseName,
String normMethod,
String chrSmaller,
ArrayList<DesiredChrContact> contactsToCheck,
String resolution,
double minValue,
TextFile outWriter)
throws IOException {
// ReadIn normalization chr1
TextFile inputNormChr1 =
new TextFile(
baseName + "\\chr" + chrSmaller + "_" + resolution + "." + normMethod, TextFile.R);
ArrayList<String> normFactorSmallerChr = inputNormChr1.readAsArrayList();
inputNormChr1.close();
// System.out.println("Done reading norm factor 1");
if (!Gpio.exists(fileToRead + ".sorted")) {
umcg.genetica.io.chrContacts.SortIntraChrContacts.readNonSortedWriteSorted(
fileToRead, fileToRead + ".sorted");
}
int numberToBeMatched = 0;
LineIterator it = FileUtils.lineIterator(new File(fileToRead + ".sorted"), "UTF-8");
try {
while (it.hasNext()) {
String[] parts = StringUtils.split(it.nextLine(), '\t');
int posChr1 = org.apache.commons.lang.math.NumberUtils.createInteger(parts[0]);
int posChr2 = org.apache.commons.lang.math.NumberUtils.createInteger(parts[1]);
while (numberToBeMatched < contactsToCheck.size()) {
if (posChr1 < contactsToCheck.get(numberToBeMatched).getChrLocationSmaller()) {
break;
} else if (posChr1 == contactsToCheck.get(numberToBeMatched).getChrLocationSmaller()) {
if (posChr2 < contactsToCheck.get(numberToBeMatched).getChrLocationLarger()) {
break;
}
if (posChr2 == contactsToCheck.get(numberToBeMatched).getChrLocationLarger()) {
String factor1Base =
normFactorSmallerChr.get((posChr1 / getNumericResolution(resolution)) + 1);
String factor2Base =
normFactorSmallerChr.get((posChr2 / getNumericResolution(resolution)) + 1);
double factor1;
double factor2;
if (StringUtils.isNumeric(factor1Base) && StringUtils.isNumeric(factor2Base)) {
factor1 = org.apache.commons.lang.math.NumberUtils.createDouble(factor1Base);
factor2 = org.apache.commons.lang.math.NumberUtils.createDouble(factor2Base);
double contact =
org.apache.commons.lang.math.NumberUtils.createDouble(parts[2])
/ (factor1 * factor2);
if (contact >= minValue) {
outWriter.writeln(
contactsToCheck.get(numberToBeMatched).getSnpName()
+ "\t"
+ contactsToCheck.get(numberToBeMatched).getProbeName()
+ "\t"
+ posChr1
+ "\t"
+ posChr2
+ "\tContact\t"
+ contact
+ "\t"
+ org.apache.commons.lang.math.NumberUtils.createDouble(parts[2]));
numberToBeMatched++;
} else {
outWriter.writeln(
contactsToCheck.get(numberToBeMatched).getSnpName()
+ "\t"
+ contactsToCheck.get(numberToBeMatched).getProbeName()
+ "\t"
+ posChr1
+ "\t"
+ posChr2
+ "\t-\t-\t-");
numberToBeMatched++;
}
} else {
System.out.println("Error in files.");
numberToBeMatched++;
}
} else if (posChr2 > contactsToCheck.get(numberToBeMatched).getChrLocationLarger()) {
outWriter.writeln(
contactsToCheck.get(numberToBeMatched).getSnpName()
+ "\t"
+ contactsToCheck.get(numberToBeMatched).getProbeName()
+ "\t"
+ posChr1
+ "\t"
+ posChr2
+ "\t-\t-\t-");
numberToBeMatched++;
}
} else if (posChr1 > contactsToCheck.get(numberToBeMatched).getChrLocationSmaller()) {
outWriter.writeln(
contactsToCheck.get(numberToBeMatched).getSnpName()
+ "\t"
+ contactsToCheck.get(numberToBeMatched).getProbeName()
+ "\t"
+ posChr1
+ "\t"
+ posChr2
+ "\t-\t-\t-");
numberToBeMatched++;
}
}
}
} finally {
LineIterator.closeQuietly(it);
}
}