Support paired reads ("_r[1|2]" or "_R[1|2]") in k-mer counter

ctlab · Mar 25, 2022 · d3bb960 · d3bb960
1 parent de718c6
commit d3bb960
Show file tree

Hide file tree

Showing 2 changed files with 61 additions and 19 deletions.
diff --git a/src/tools/KmersCounterForManyFilesMain.java b/src/tools/KmersCounterForManyFilesMain.java
@@ -1,8 +1,8 @@
 package tools;
 
 import io.IOUtils;
+import ru.ifmo.genetics.io.ReadersUtils;
 import ru.ifmo.genetics.statistics.Timer;
-import ru.ifmo.genetics.utils.TextUtils;
 import ru.ifmo.genetics.utils.tool.ExecutionFailedException;
 import ru.ifmo.genetics.utils.tool.Parameter;
 import ru.ifmo.genetics.utils.tool.Tool;
@@ -12,8 +12,10 @@
 import ru.ifmo.genetics.utils.tool.values.InMemoryValue;
 import ru.ifmo.genetics.utils.tool.values.InValue;
 
-import java.io.*;
+import java.io.File;
+import java.io.IOException;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.List;
 
 public class KmersCounterForManyFilesMain extends Tool {
@@ -64,22 +66,47 @@ public class KmersCounterForManyFilesMain extends Tool {
     private Timer t;
 
     @Override
-    protected void runImpl() throws ExecutionFailedException {
+    protected void runImpl() throws ExecutionFailedException, IOException {
         t = new Timer();
         counters.clear();
 
-        for (File f : inputFiles.get()) {
-            KmersCounterMain counter = new KmersCounterMain();
-            counter.workDir.set(workDir.append("sub-counter"));
-            counter.k.set(k);
-            counter.inputFiles.set(new File[]{f});
-            counter.maximalBadFrequency.set(maximalBadFrequency);
-            counter.outputDir.set(outputDir);
-            counter.statsDir.set(statsDir);
-
-            addStep(counter);
-            counters.add(counter);
+        File[] files = inputFiles.get();
+        Arrays.sort(files);
+        List<String> names = new ArrayList<String>();
+        for (File f : files) {
+            names.add(ReadersUtils.readDnaLazy(f).name());
         }
+
+        int i=0;
+        while (i < files.length) {
+            if ((names.get(i).endsWith("_r1") && i+1<files.length && names.get(i+1).endsWith("_r2")) ||
+                    (names.get(i).endsWith("_R1") && i+1<files.length && names.get(i+1).endsWith("_R2"))) {
+                KmersCounterMain counter = new KmersCounterMain();
+                counter.workDir.set(workDir.append("sub-counter"));
+                counter.k.set(k);
+                counter.inputFiles.set(new File[]{files[i], files[i+1]});
+                counter.maximalBadFrequency.set(maximalBadFrequency);
+                counter.outputDir.set(outputDir);
+                counter.statsDir.set(statsDir);
+
+                addStep(counter);
+                counters.add(counter);
+                i += 2;
+            } else {
+                KmersCounterMain counter = new KmersCounterMain();
+                counter.workDir.set(workDir.append("sub-counter"));
+                counter.k.set(k);
+                counter.inputFiles.set(new File[]{files[i]});
+                counter.maximalBadFrequency.set(maximalBadFrequency);
+                counter.outputDir.set(outputDir);
+                counter.statsDir.set(statsDir);
+
+                addStep(counter);
+                counters.add(counter);
+                i += 1;
+            }
+        }
+
     }
 
     @Override

diff --git a/src/tools/KmersCounterMain.java b/src/tools/KmersCounterMain.java
@@ -1,20 +1,19 @@
 package tools;
 
 import io.IOUtils;
-import ru.ifmo.genetics.dna.kmers.ShortKmerIteratorFactory;
 import ru.ifmo.genetics.io.ReadersUtils;
 import ru.ifmo.genetics.statistics.Timer;
 import ru.ifmo.genetics.structures.map.BigLong2ShortHashMap;
 import ru.ifmo.genetics.utils.Misc;
-import ru.ifmo.genetics.utils.tool.*;
+import ru.ifmo.genetics.utils.NumUtils;
+import ru.ifmo.genetics.utils.tool.ExecutionFailedException;
 import ru.ifmo.genetics.utils.tool.Parameter;
 import ru.ifmo.genetics.utils.tool.Tool;
 import ru.ifmo.genetics.utils.tool.inputParameterBuilder.FileMVParameterBuilder;
 import ru.ifmo.genetics.utils.tool.inputParameterBuilder.FileParameterBuilder;
 import ru.ifmo.genetics.utils.tool.inputParameterBuilder.IntParameterBuilder;
 import ru.ifmo.genetics.utils.tool.values.InMemoryValue;
 import ru.ifmo.genetics.utils.tool.values.InValue;
-import ru.ifmo.genetics.utils.NumUtils;
 
 import java.io.File;
 import java.io.IOException;
@@ -89,8 +88,7 @@ protected void runImpl() throws ExecutionFailedException, IOException {
             stDir.mkdirs();
         }
 
-        String name = ReadersUtils.readDnaLazy(inputFiles.get()[0]).name()
-                            + (inputFiles.get().length > 1 ? "+" : "");
+        String name = getName();
         File outFile = new File(outDir, name + ".kmers.bin");
         File stFile = new File(stDir, name + ".stat.txt");
 
@@ -121,6 +119,23 @@ protected void runImpl() throws ExecutionFailedException, IOException {
         resultingKmerFilesPr.set(outFile);
     }
 
+    private String getName() throws IOException {
+        if (inputFiles.get().length == 2) {
+            String name1 = ReadersUtils.readDnaLazy(inputFiles.get()[0]).name();
+            String name2 = ReadersUtils.readDnaLazy(inputFiles.get()[1]).name();
+            if ((name1.endsWith("_r1") && name2.endsWith("_r2")) ||
+                (name1.endsWith("_R1") && name2.endsWith("_R2"))) {
+                return name1.substring(0, name1.length()-3);
+            } else {
+                return name1 + "+";
+            }
+
+        } else {
+            return ReadersUtils.readDnaLazy(inputFiles.get()[0]).name()
+                    + (inputFiles.get().length > 1 ? "+" : "");
+        }
+    }
+
     @Override
     protected void cleanImpl() {
     }