Export best n hits, solves #12

src/main/java/genepi/haplogrep/main/Haplogrep.java

@@ -33,8 +33,9 @@ public void createParameters() {
 		addFlag("rsrs", "use RSRS Version");
 		addFlag("extend-report", "add flag for a extended final output");
 		addFlag("chip", "VCF data from a genotype chip");
-		addOptionalParameter("metric", "specifiy other metric (hamming or jaccard)", Tool.STRING);
+		addOptionalParameter("metric", "specifiy other metric (hamming or jaccard) than default (kulczynski)", Tool.STRING);
 		addFlag("lineage", "export lineage information");
+		addOptionalParameter("hits", "calculate best n hits", Tool.STRING);
 	}
 
 	@Override
@@ -55,6 +56,7 @@ public int run() {
 		String tree = (String) getValue("phylotree");
 		String format = (String) getValue("format");
 		String metric = (String) getValue("metric");
+		String hits = (String) getValue("hits");
 
 		boolean extended = isFlagSet("extend-report");
 
@@ -78,6 +80,10 @@ public int run() {
 			tree = "17";
 		}
 
+		if (hits == null) {
+			hits = "1";
+		}
+
 		File input = new File(in);
 
 		if (!input.exists()) {
@@ -151,15 +157,18 @@ else if (format.equals("fasta")) {
 
 					SampleFile newSampleFile = new SampleFile(lines);
 
-					HgClassifier.run(newSampleFile, phylotree, fluctrates, metric);
-
+					HgClassifier classifier = new HgClassifier();
+
+					classifier.run(newSampleFile, phylotree, fluctrates, metric, Integer.valueOf(hits));
+
 					ArrayList<TestSample> samples = newSampleFile.getTestSamples();
 
 					ExportUtils.createReport(samples, out, extended);
 
 					if (lineage) {
 						ExportUtils.calcLineage(samples, out);
 					}
+
 				}
 
 			} else {
@@ -182,9 +191,8 @@ public static void main(String[] args) throws IOException {
 
 		Haplogrep haplogrep = new Haplogrep(args);
 
-		//haplogrep = new Haplogrep(
-		//		new String[] { "--in", "/home/seb/Desktop/1000G_BAQ.vcf.gz",
-		//				"--out", "/home/seb/Desktop/test.txt", "--format", "vcf", "--extend-report" });
+		//haplogrep = new Haplogrep(new String[] { "--in", "test-data/vcf/HG00097.vcf", "--out",
+		//		"test-data/test.txt", "--format", "vcf","--hits", "5"});
 
 		haplogrep.start();
 

src/main/java/genepi/haplogrep/util/HgClassifier.java

@@ -14,9 +14,13 @@
 import search.ranking.RankingMethod;
 
 public class HgClassifier {
-
-	public static void run(SampleFile newSampleFile, String phyloTree, String fluctrates, String metric)
-			throws JDOMException, IOException, InvalidRangeException {
+
+	public void run(SampleFile newSampleFile, String phyloTree, String fluctrates, String metric) throws InvalidRangeException, JDOMException, IOException {
+		run(newSampleFile, phyloTree, fluctrates, metric,1);
+	}
+
+	public void run(SampleFile newSampleFile, String phyloTree, String fluctrates, String metric,
+			int amountResults) throws JDOMException, IOException, InvalidRangeException {
 
 		Phylotree phylotree = PhylotreeManager.getInstance().getPhylotree(phyloTree, fluctrates);
 
@@ -25,19 +29,19 @@ public static void run(SampleFile newSampleFile, String phyloTree, String fluctr
 		switch (metric) {
 
 		case "kulczynski":
-			newRanker = new KulczynskiRanking(1);
+			newRanker = new KulczynskiRanking(amountResults);
 			break;
 
 		case "hamming":
-			newRanker = new HammingRanking(1);
+			newRanker = new HammingRanking(amountResults);
 			break;
 
 		case "jaccard":
-			newRanker = new JaccardRanking(1);
+			newRanker = new JaccardRanking(amountResults);
 			break;
 
 		default:
-			newRanker = new KulczynskiRanking(1);
+			newRanker = new KulczynskiRanking(amountResults);
 
 		}
 

src/test/java/genepi/haplogrep/HaplogrepCmdTest.java

@@ -0,0 +1,71 @@
+package genepi.haplogrep;
+
+import static org.junit.Assert.assertEquals;
+
+import java.io.File;
+import java.util.ArrayList;
+import java.util.HashMap;
+
+import org.junit.Test;
+
+import contamination.objects.Sample;
+import core.SampleFile;
+import genepi.haplogrep.util.HgClassifier;
+import genepi.io.FileUtil;
+import genepi.io.table.reader.CsvTableReader;
+import importer.VcfImporter;
+import util.ExportUtils;
+
+public class HaplogrepCmdTest {
+
+	@Test
+	public void HaplogrepCmdTest() throws Exception {
+
+		String file = "test-data/vcf/HG00097.vcf";
+		String phylo = "data/phylotree/phylotree17.xml";
+		String weights = "data/weights/weights17.txt";
+		String out = "test-data/out.txt";
+		VcfImporter impvcf = new VcfImporter();
+
+		HashMap<String, Sample> samples = impvcf.load(new File(file), false);
+		ArrayList<String> lines = ExportUtils.samplesMapToHsd(samples);
+		SampleFile newSampleFile = new SampleFile(lines);
+
+		HgClassifier classifier = new HgClassifier();
+
+		classifier.run(newSampleFile, phylo, weights, "kulczynski", 1);
+
+		ExportUtils.createReport(newSampleFile.getTestSamples(), out, false);
+
+		CsvTableReader reader = new CsvTableReader(out, '\t');
+
+		int count = 0;
+		while (reader.next()) {
+			count++;
+			String hg = reader.getString("Haplogroup");
+			double quality = reader.getDouble("Quality");
+			assertEquals("T2f1a1", hg);
+			assertEquals(quality, quality,0.0);
+		}
+
+		assertEquals(1, count);
+
+
+		classifier.run(newSampleFile, phylo, weights, "kulczynski", 10);
+
+		ExportUtils.createReport(newSampleFile.getTestSamples(), out, false);
+
+		reader = new CsvTableReader(out, '\t');
+
+		count = 0;
+		while (reader.next()) {
+			count++;
+		}
+
+		assertEquals(10, count);
+
+		FileUtil.deleteFile(out);
+
+	}
+
+}