Merge pull request molgenis#561 from Bonder-MJ/master

Minor updates

BinaryMetaAnalyzer/src/main/java/nl/umcg/westrah/binarymetaanalyzer/BinaryMetaAnalysisSettings.java

@@ -27,7 +27,7 @@ public String toString() {
 				"useAbsoluteZscore\t" + useAbsoluteZscore + "\n" +
 				"finalEQTLBufferMaxLength\t" + finalEQTLBufferMaxLength + "\n" +
 				"includeSNPsWithoutProperMapping\t" + includeSNPsWithoutProperMapping + "\n" +
-				"includeProbesWithoutProperMapping\t" + includeProbesWithoutProperMapping +
+				"includeProbesWithoutProperMapping\t" + includeProbesWithoutProperMapping + "\n" +
 				"analysisType\t" + analysisType + "\n" +
 				"cisdistance\t" + cisdistance + "\n" +
 				"transdistance\t" + transdistance + "\n" +
@@ -45,6 +45,7 @@ public String toString() {
 				"snpselection\t'" + snpselection + '\'' + "\n" +
 				"snpprobeselection\t'" + snpprobeselection + '\'' + "\n" +
 				"snpAnnotationFile\t'" + snpAnnotationFile + '\'' + "\n" +
+				"probeRescaling\t" + rescalingOfSampleSize + "\n" +
 				"minimalNumberOfDatasets\t" + minimalNumberOfDatasets + "\n" +
 				"debug\t" + debug + "\n" +
 				"fullpermutationoutput\t" + fullpermutationoutput + "\n" +
@@ -249,7 +250,6 @@ public void parse(String settings, String texttoreplace, String replacetextwith)
 				}
 				i++;
 			}
-
 			summarize();
 			// parse datasets
 		} catch (ConfigurationException e) {

BinaryMetaAnalyzer/src/main/java/nl/umcg/westrah/binarymetaanalyzer/BinaryMetaAnalysisTask.java

@@ -392,14 +392,21 @@ public Triple<ArrayList<QTL>, String, String> call() {
 
 					if (settings.getRescalingOfSampleSize()) {
 						weights = new double[sampleSizes.length];
-						for (int d = 0; d < datasetZScores.length; d++) {
-							Double rescaleValue = datasets[d].getFeatureOccuranceScaleMap().get(t.getPlatformIds()[d]);
-							if (!Double.isFinite(rescaleValue)) {
-								System.out.println("Warning for feature: " + t.getPlatformIds()[d] + " no rescale value set for: " + datasets[d].getName() + "\n Defaulted to weight of 1.");
+						for(int d=0; d<datasetZScores.length;d++){
+							if(debug){
+								System.out.println("Dataset: "+d);
+								System.out.println("Dataset name: "+datasets[d].getName());
+								System.out.println("Dataset platformId: "+t.getPlatformIds()[0]);
+								System.out.println("Dataset rescaling size: "+datasets[d].getFeatureOccuranceScaleMap().size());
+								System.out.println("Dataset rescaling contains key: "+datasets[d].getFeatureOccuranceScaleMap().containsKey(t.getPlatformIds()[0]));
+							}
+							Double rescaleValue = datasets[d].getFeatureOccuranceScaleMap().get(t.getPlatformIds()[0]);
+							if(rescaleValue==null || !Double.isFinite(rescaleValue)){
+								System.out.println("Warning for feature: "+ t.getPlatformIds()[0] + " no rescale value set for: " + datasets[d].getName()+"\n Defaulted to weight of 1.");
 								rescaleValue = 1.0d;
 							}
-							if (debug) {
-								System.out.println("Weight for: " + t.getPlatformIds()[d] + " is: " + rescaleValue);
+							if(debug){
+								System.out.println("Weight for: "+t.getPlatformIds()[0]+" is: "+rescaleValue);
 							}
 							weights[d] = rescaleValue;
 						}
@@ -411,7 +418,7 @@ public Triple<ArrayList<QTL>, String, String> call() {
 						if (!settings.getRescalingOfSampleSize()) {
 							metaAnalysisZ = ZScores.getWeightedZ(datasetZScores, sampleSizes);
 						} else {
-							metaAnalysisZ = ZScores.getWeightedZ(datasetZScores, sampleSizes, weights);
+							metaAnalysisZ = ZScores.getDatasetSizeWeightedZ(datasetZScores, sampleSizes, weights);
 						}
 
 						if (debug && traitList[probe].getMetaTraitName().equals("ENSG00000132465")) {

GeneticRiskScoreCalculator/pom.xml

@@ -93,7 +93,7 @@
                             <classpathPrefix>lib/</classpathPrefix>
                             <!-- Specifies the main class of the application -->
                             <mainClass>nl.systemsgenetics.simplegeneticriskscorecalculator.Main</mainClass>
-<!--                            <mainClass>nl.systemsgenetics.benchmark.Benchmark</mainClass>-->
+                            <!--<mainClass>nl.systemsgenetics.benchmark.Benchmark</mainClass>-->
                         </manifest>
                     </archive>
                 </configuration>

GeneticRiskScoreCalculator/src/main/java/nl/systemsgenetics/simplegeneticriskscorecalculator/Main.java

@@ -54,7 +54,7 @@ public static void main(String[] args) {
 		Option FileOut = OptionBuilder.withArgName("path").hasArg().withDescription("Location (folder) for the output.").withLongOpt("OutputFolder").create("o");
 		Option GenotypeTypeIn = OptionBuilder.withArgName("type").hasArg().withDescription("Type of reference data.").withLongOpt("GenotypeType").create("gt");
 		Option GenotypeIn = OptionBuilder.withArgName("path").hasArg().withDescription("Location for the reference data.").withLongOpt("GenotypeLocation").create("gi");
-		Option InFolder = OptionBuilder.withArgName("path").hasArg().withDescription("Location of the folder with genetic risk score information.").withLongOpt("input").create("i");
+		Option InFolder = OptionBuilder.withArgName("path").hasArg().withDescription("Location of the file/folder with genetic risk score information.").withLongOpt("input").create("i");
 		Option rSquared = OptionBuilder.withArgName("double").hasArg().withDescription("R2 for pruning.").withLongOpt("rSquared").create("r");
 		Option pValueThreshold = OptionBuilder.withArgName("double").hasArg().withDescription("P-value thresholds for genetic risk score inclusion, colon separated should be ordered from most stringent to least stringent.").withLongOpt("pValue").create("p");
 		Option WindowSize = OptionBuilder.withArgName("double").hasArg().withDescription("Window size for pruning, if given two window-sizes (colon separated), a two step window approach is used.").withLongOpt("wSize").create("w");
@@ -205,11 +205,17 @@ private static THashMap<String, THashMap<String, THashMap<String, ArrayList<Risk
 
 		File riskFileFolder = new File(riskFolder);
 		if (!riskFileFolder.exists()) {
-			System.out.println("Warning: input risk folder does not exists:\n");
+			System.out.println("Warning: input risk file / folder does not exists:\n");
 			System.out.println(riskFileFolder);
 			System.exit(-1);
 		}
-		File[] riskFiles = riskFileFolder.listFiles();
+		File[] riskFiles;
+		if(riskFileFolder.isDirectory()){
+			riskFiles = riskFileFolder.listFiles();
+		} else {
+			riskFiles = new File[1];
+			riskFiles[0] = riskFileFolder;
+		}
 
 
 		IntStream.range(0, riskFiles.length).parallel().forEach(fi -> {

Genotype-IO/src/test/java/org/molgenis/genotype/bgen/BgenGenotypeDataTest.java

@@ -441,3 +441,4 @@ public void TestBgenGenotypeDataBgenix() throws URISyntaxException, IOException
 		}
 	}
 }
+

genetica-libraries/src/main/java/umcg/genetica/math/stats/ZScores.java

@@ -105,6 +105,30 @@ public static double getWeightedZ(float[] zScores, int[] sampleSizes, double[] w
 		return weightedZ;
 	}
 
+	/**
+	 * Calculates a weighted Z-score according to Whitlock's paper:
+	 * http://www.ncbi.nlm.nih.gov/pubmed/16135132 Square root of the sample
+	 * size is used as the weight for each test.
+	 *
+	 * @param zScores     Z-scores from individual tests
+	 * @param sampleSizes sample sizes of these tests
+	 * @return
+	 */
+	public static double getDatasetSizeWeightedZ(float[] zScores, int[] sampleSizes, double[] weights) {
+		if (zScores.length != sampleSizes.length) {
+			throw new IllegalArgumentException("Zscores and sample sizes should have same length!");
+		}
+		double weightedZ = 0;
+		double sampleSizeSum = 0;
+		for (int j = 0; j < zScores.length; j++) {
+			if (!Double.isNaN(zScores[j])) {
+				weightedZ += Math.sqrt(sampleSizes[j]* weights[j]) * zScores[j];
+				sampleSizeSum += (sampleSizes[j] * weights[j]);
+			}
+		}
+		weightedZ /= Math.sqrt(sampleSizeSum);
+		return weightedZ;
+	}
 	/**
 	 * Calculates a weighted Z-score according to Whitlock's paper:
 	 * http://www.ncbi.nlm.nih.gov/pubmed/16135132 Square root of the sample