formats: improve samtools/flagstat support, #187

biodata-formats/src/main/java/org/opencb/biodata/formats/alignment/samtools/SamtoolsFlagstats.java

@@ -22,6 +22,10 @@ public class SamtoolsFlagstats {
      */
     private int secondaryAlignments;
 
+    /**
+     * Total number of reads which are supplementary (0x800 bit set)
+     */
+    private int supplementary;
     /**
      * Total number of reads which are duplicates (0x400 bit set)
      */
@@ -32,6 +36,17 @@ public class SamtoolsFlagstats {
      */
     private int pairedInSequencing;
 
+    /**
+     * Total number of reads with both 0x1 and 0x40 bits set
+     */
+    private int read1;
+
+
+    /**
+     * Total number reads with both 0x1 and 0x80 bits set
+     */
+    private int read2;
+
     /**
      * Total number of reads which are properly paired (both 0x1 and 0x2 bits set and 0x4 bit not set)
      */
@@ -70,8 +85,11 @@ public String toString() {
         sb.append(", totalQcPassed=").append(totalQcPassed);
         sb.append(", mapped=").append(mapped);
         sb.append(", secondaryAlignments=").append(secondaryAlignments);
+        sb.append(", supplementary=").append(supplementary);
         sb.append(", duplicates=").append(duplicates);
         sb.append(", pairedInSequencing=").append(pairedInSequencing);
+        sb.append(", read1=").append(read1);
+        sb.append(", read2=").append(read2);
         sb.append(", properlyPaired=").append(properlyPaired);
         sb.append(", selfAndMateMapped=").append(selfAndMateMapped);
         sb.append(", singletons=").append(singletons);
@@ -117,6 +135,15 @@ public SamtoolsFlagstats setSecondaryAlignments(int secondaryAlignments) {
         return this;
     }
 
+    public int getSupplementary() {
+        return supplementary;
+    }
+
+    public SamtoolsFlagstats setSupplementary(int supplementary) {
+        this.supplementary = supplementary;
+        return this;
+    }
+
     public int getDuplicates() {
         return duplicates;
     }
@@ -135,6 +162,24 @@ public SamtoolsFlagstats setPairedInSequencing(int pairedInSequencing) {
         return this;
     }
 
+    public int getRead1() {
+        return read1;
+    }
+
+    public SamtoolsFlagstats setRead1(int read1) {
+        this.read1 = read1;
+        return this;
+    }
+
+    public int getRead2() {
+        return read2;
+    }
+
+    public SamtoolsFlagstats setRead2(int read2) {
+        this.read2 = read2;
+        return this;
+    }
+
     public int getProperlyPaired() {
         return properlyPaired;
     }

biodata-formats/src/main/java/org/opencb/biodata/formats/alignment/samtools/io/SamtoolsFlagstatsParser.java

@@ -0,0 +1,57 @@
+package org.opencb.biodata.formats.alignment.samtools.io;
+
+import org.opencb.biodata.formats.alignment.samtools.SamtoolsFlagstats;
+
+import java.io.BufferedReader;
+import java.io.FileReader;
+import java.io.IOException;
+import java.nio.file.Path;
+
+public class SamtoolsFlagstatsParser {
+
+    public static SamtoolsFlagstats parse(Path path) throws IOException {
+
+        SamtoolsFlagstats flagstats = new SamtoolsFlagstats();
+
+        FileReader fr = new FileReader(path.toFile());
+        BufferedReader br = new BufferedReader(fr);
+
+        String line;
+
+        while ((line = br.readLine()) != null) {
+            String[] splits = line.split(" ");
+            int passed = Integer.parseInt(splits[0]);
+            int failed = Integer.parseInt(splits[2]);
+            if (line.contains("QC-passed")) {
+                flagstats.setTotalQcPassed(passed);
+                flagstats.setTotalReads(passed + failed);
+            } else if (line.contains("secondary")) {
+                flagstats.setSecondaryAlignments(passed);
+            } else if (line.contains("supplementary")) {
+                flagstats.setSupplementary(passed);
+            } else if (line.contains("duplicates")) {
+                flagstats.setDuplicates(passed);
+            } else if (line.contains("mapped")) {
+                flagstats.setMapped(passed);
+            } else if (line.contains("paired in sequencing")) {
+                flagstats.setPairedInSequencing(passed);
+            } else if (line.contains("read1")) {
+                flagstats.setRead1(passed);
+            } else if (line.contains("read2")) {
+                flagstats.setRead2(passed);
+            } else if (line.contains("properly paired")) {
+                flagstats.setProperlyPaired(passed);
+            } else if (line.contains("with itself and mate mapped")) {
+                flagstats.setSelfAndMateMapped(passed);
+            } else if (line.contains("singletons")) {
+                flagstats.setSingletons(passed);
+            } else if (line.contains("mapQ>=5")) {
+                flagstats.setDiffChrMapQ5(passed);
+            } else if (line.contains("with mate mapped")) {
+                flagstats.setMateMappedToDiffChr(passed);
+            }
+        }
+
+        return flagstats;
+    }
+}