Features/filtering (#17)

* Add missing setter methods

* Add include-scores option

src/main/java/genepi/riskscore/commands/ApplyScoreCommand.java

@@ -69,6 +69,10 @@ public class ApplyScoreCommand implements Callable<Integer> {
 	@Option(names = { "--samples" }, description = "Include only samples from this file", required = false)
 	String includeSamplesFilename = null;
 
+	@Option(names = { "--includeScores",
+			"--include-scores" }, description = "Include only scores from this file. Works only with score collections.", required = false)
+	String includeScoresFilename = null;
+
 	@Option(names = { "--report-json", "--info" }, description = "Write statistics to json file", required = false)
 	String reportJson = null;
 
@@ -202,6 +206,7 @@ public Integer call() throws Exception {
 			}
 			task.setIncludeVariantFilename(includeVariantFilename);
 			task.setIncludeSamplesFilename(includeSamplesFilename);
+			task.setIncludeScoreFilename(includeScoresFilename);
 			task.setOutput(taskPrefix + ".scores.txt");
 			task.setFixStrandFlips(fixStrandFlips);
 			task.setRemoveAmbiguous(removeAmbiguous);

src/main/java/genepi/riskscore/commands/FilterMetaCommand.java

@@ -70,5 +70,19 @@ private boolean accept(MetaFile.MetaScore score, String category, String populat
 		return true;
 	}
 
+	public void setMeta(String meta) {
+		this.meta = meta;
+	}
+
+	public void setCategory(String category) {
+		this.category = category;
+	}
 
+	public void setPopulation(String population) {
+		this.population = population;
+	}
+
+	public void setOut(String out) {
+		this.out = out;
+	}
 }

src/main/java/genepi/riskscore/io/scores/MergedRiskScoreCollection.java

@@ -1,5 +1,6 @@
 package genepi.riskscore.io.scores;
 
+import genepi.io.text.LineReader;
 import genepi.riskscore.io.Chunk;
 import genepi.riskscore.io.csv.CsvWithHeaderTableReader;
 import genepi.riskscore.io.csv.TabixTableReader;
@@ -23,6 +24,8 @@ public class MergedRiskScoreCollection implements IRiskScoreCollection {
 
 	private String filename;
 
+	private String includeScoresFilename;
+
 	private boolean verbose = false;
 
 	private Map<String, MergedVariant> variantsIndex = new HashMap<String, MergedVariant>();
@@ -52,8 +55,9 @@ public class MergedRiskScoreCollection implements IRiskScoreCollection {
 		COLUMNS.add(COLUMN_OTHER_ALLELE);
 	}
 
-	public MergedRiskScoreCollection(String filename) {
+	public MergedRiskScoreCollection(String filename, String includeScoresFilename) {
 		this.filename = filename;
+		this.includeScoresFilename = includeScoresFilename;
 	}
 
 	@Override
@@ -74,6 +78,19 @@ public String getVersion() {
 	@Override
 	public void buildIndex(String chromosome, Chunk chunk, String dbsnp, String proxies) throws Exception {
 
+		List<String> includedScores = new Vector<String>();
+
+		if (includeScoresFilename != null) {
+			LineReader lineReader = new LineReader(includeScoresFilename);
+			while(lineReader.next()){
+				String score = lineReader.get().trim();
+				if (!score.isEmpty()) {
+					includedScores.add(score);
+				}
+			}
+			lineReader.close();
+		}
+
 		String metaFilename = filename + META_EXTENSION;
 		File metaFile = new File(metaFilename);
 		if (!metaFile.exists()){
@@ -95,12 +112,25 @@ public void buildIndex(String chromosome, Chunk chunk, String dbsnp, String prox
 		TabixTableReader reader = new TabixTableReader(filename, chromosome, chunk.getStart(), chunk.getEnd());
 		String[] columns = reader.getColumns();
 
-		numberRiskScores = columns.length - COLUMNS.size();
+		Map<String, Integer> scoreToColumn = new HashMap<String, Integer>();
+		summaries = new RiskScoreSummary[numberRiskScores];
+		for (int i = 0; i < columns.length; i++){
+			String column = columns[i];
+			if (COLUMNS.contains(column)){
+				continue;
+			}
+			if (!includedScores.isEmpty() && !includedScores.contains(column)){
+				continue;
+			}
+			scoreToColumn.put(column, i);
+		}
+
 
+		numberRiskScores = scoreToColumn.size();
 		summaries = new RiskScoreSummary[numberRiskScores];
 		int index = 0;
 		for (String column: columns){
-			if (COLUMNS.contains(column)){
+			if (!scoreToColumn.containsKey(column)){
 				continue;
 			}
 			summaries[index] = new RiskScoreSummary(column);
@@ -129,7 +159,7 @@ public void buildIndex(String chromosome, Chunk chunk, String dbsnp, String prox
 			}
 			index = 0;
 			for (String column: columns){
-				if (COLUMNS.contains(column)){
+				if (!scoreToColumn.containsKey(column)){
 					continue;
 				}
 				if (reader.getString(column).equals("") || reader.getString(column).equals(".")){

src/main/java/genepi/riskscore/tasks/ApplyScoreTask.java

@@ -62,6 +62,8 @@ public class ApplyScoreTask implements ITaskRunnable {
 
 	private String proxies;
 
+	private String includeScoreFilename = null;
+
 	private boolean fixStrandFlips = false;
 
 	private boolean removeAmbiguous = false;
@@ -114,6 +116,10 @@ public void setIncludeSamplesFilename(String includeSamplesFilename) {
 		this.includeSamplesFilename = includeSamplesFilename;
 	}
 
+	public void setIncludeScoreFilename(String includeScoreFilename) {
+		this.includeScoreFilename = includeScoreFilename;
+	}
+
 	public void setGenotypeFormat(String genotypeFormat) {
 		this.genotypeFormat = genotypeFormat;
 	}
@@ -172,7 +178,7 @@ public void run(ITaskMonitor monitor) throws Exception {
 			//TODO: move to factory
 			if (riskScoreFilenames.length == 1 && new File(riskScoreFilenames[0]).exists() &&
 					RiskScoreFormatFactory.readHeader(riskScoreFilenames[0]).startsWith(MergedRiskScoreCollection.HEADER)) {
-				collection = new MergedRiskScoreCollection(riskScoreFilenames[0]);
+				collection = new MergedRiskScoreCollection(riskScoreFilenames[0], includeScoreFilename);
 			} else {
 				collection = new RiskScoreCollection(riskScoreFilenames, formats);
 			}