Update EnvironmentFinderMultiMain.java

src/algo/AssemblerCalculator.java

@@ -0,0 +1,104 @@
+package algo;
+
+import org.apache.log4j.Logger;
+
+import java.io.BufferedReader;
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.InputStreamReader;
+
+/**
+ * Created by -- on 13.02.2018.
+ */
+public class AssemblerCalculator implements Runnable {
+    private final String assembler;
+    private final String assemblerPath;
+    private final String outputPrefix;
+    private final int readsNumber;
+    private final Logger logger;
+
+    public AssemblerCalculator(String assembler, String assemblerPath, String outputPrefix, int readsNumber, Logger logger) {
+        this.assembler = assembler;
+        this.assemblerPath = assemblerPath;
+        this.outputPrefix = outputPrefix;
+        this.readsNumber = readsNumber;
+        this.logger = logger;
+    }
+
+    private void runAssembler() {
+        if (assembler.equals("spades")) {
+            try {
+                String[] commandA = {"python", assemblerPath + "/spades.py", "--12",
+                        outputPrefix + "cutReads" + readsNumber + ".fastq", "-o", outputPrefix + "out_spades" + readsNumber};
+                String[] commandR = {"mv", outputPrefix + "out_spades" + readsNumber + "/" + "contigs.fasta",
+                        outputPrefix + "contigs" + readsNumber + ".fasta"};
+                ProcessBuilder procBuilder = new ProcessBuilder(commandA);
+                ProcessBuilder procBuilderR = new ProcessBuilder(commandR);
+
+                procBuilder.redirectErrorStream(true);
+                Process process = procBuilder.start();
+                InputStream stdout = process.getInputStream();
+                InputStreamReader isrStdout = new InputStreamReader(stdout);
+                BufferedReader brStdout = new BufferedReader(isrStdout);
+                String line;
+                while ((line = brStdout.readLine()) != null) {
+                    logger.info(line);
+                }
+                // or better take assembly_graph.gfa?
+
+                procBuilderR.redirectErrorStream(true);
+                Process processB = procBuilderR.start();
+                InputStream stdoutB = processB.getInputStream();
+                InputStreamReader isrStdoutB = new InputStreamReader(stdoutB);
+                BufferedReader brStdoutB = new BufferedReader(isrStdoutB);
+                String lineB;
+                while ((lineB = brStdoutB.readLine()) != null) {
+                    logger.info(lineB);
+                }
+            } catch (IOException e) {
+                e.printStackTrace();
+                logger.info(e.getMessage());
+            }
+        }
+        if (assembler.equals("megahit")) {
+            try {
+                String[] commandA = {assemblerPath + "/megahit", "--12",
+                        outputPrefix + "cutReads" + readsNumber + ".fastq", "-o", outputPrefix + "out_megahit" + readsNumber};
+                String[] commandR = {"mv", outputPrefix + "out_megahit" + readsNumber + "/" + "final.contigs.fa",
+                        outputPrefix + "contigs" + readsNumber + ".fasta"};
+                ProcessBuilder procBuilder = new ProcessBuilder(commandA);
+                ProcessBuilder procBuilderR = new ProcessBuilder(commandR);
+
+                procBuilder.redirectErrorStream(true);
+                Process process = procBuilder.start();
+                InputStream stdout = process.getInputStream();
+                InputStreamReader isrStdout = new InputStreamReader(stdout);
+                BufferedReader brStdout = new BufferedReader(isrStdout);
+
+                String line;
+                while ((line = brStdout.readLine()) != null) {
+                    logger.info(line);
+                }
+
+                procBuilderR.redirectErrorStream(true);
+                Process processB = procBuilderR.start();
+                InputStream stdoutB = processB.getInputStream();
+                InputStreamReader isrStdoutB = new InputStreamReader(stdoutB);
+                BufferedReader brStdoutB = new BufferedReader(isrStdoutB);
+                String lineB;
+                while ((lineB = brStdoutB.readLine()) != null) {
+                    logger.info(lineB);
+                }
+
+            } catch (IOException e) {
+                e.printStackTrace();
+                logger.info(e.getMessage());
+            }
+        }
+    }
+
+    @Override
+    public void run() {
+        runAssembler();
+    }
+}

src/algo/OneSequenceCalculator.java

@@ -37,7 +37,7 @@ public class OneSequenceCalculator implements Runnable {
     private boolean fail = false;
 
     public OneSequenceCalculator(String sequence, int k, int minOccurences, String outputPrefix, HashFunction hasher, BigLong2ShortHashMap reads, Logger logger, boolean bothDirections, int chunkLength, TerminationMode termMode, boolean trimPaths) {
-        this.sequence = sequence.toString();
+        this.sequence = sequence;
         this.k = k;
         this.outputPrefix = outputPrefix;
         this.minOccurences = minOccurences;
@@ -100,7 +100,7 @@ private void addToSubgraph(String kmer) {
         subgraph.put(normalizeDna(kmer), (int) reads.get(getKmerKey(kmer)));
     }
 
-    private boolean isContainedInSubgraph(String kmer) {
+    boolean isContainedInSubgraph(String kmer) {
         return subgraph.containsKey(normalizeDna(kmer));
     }
 

src/algo/ReadsFilter.java

@@ -0,0 +1,86 @@
+package algo;
+
+import java.io.File;
+import java.io.IOException;
+import java.io.PrintWriter;
+
+import org.apache.log4j.Logger;
+import ru.ifmo.genetics.dna.Dna;
+import ru.ifmo.genetics.dna.DnaQ;
+import ru.ifmo.genetics.io.ReadersUtils;
+import ru.ifmo.genetics.io.formats.QualityFormat;
+import ru.ifmo.genetics.io.sources.NamedSource;
+import ru.ifmo.genetics.utils.tool.ExecutionFailedException;
+
+public class ReadsFilter implements Runnable {
+    private OneSequenceCalculator calc;
+    private File file;
+    private final String outputPrefix;
+    private final int readsNumber;
+    private int k;
+    private int percentFiltration;
+    private final Logger logger;
+
+    public ReadsFilter(File file, OneSequenceCalculator calc, String outputPrefix, int readsNumber, int k, int percentFiltration, Logger logger) {
+        this.file = file;
+        this.calc = calc;
+        this.outputPrefix = outputPrefix;
+        this.readsNumber = readsNumber;
+        this.k = k;
+        this.percentFiltration = percentFiltration;
+        this.logger = logger;
+    }
+
+    private void reader() throws ExecutionFailedException {
+        File outputCutReads = new File(outputPrefix + "/cutReads" + readsNumber + ".fastq");
+        outputCutReads.getParentFile().mkdirs();
+        NamedSource<Dna> reader = null;
+        try {
+            PrintWriter out;
+            out = new PrintWriter(outputCutReads);
+            reader = ReadersUtils.readDnaLazy(file);
+            ReadersUtils.loadDnaQs(file);
+            QualityFormat qualF = ReadersUtils.determineQualityFormat(file);
+            NamedSource<DnaQ> quals = ReadersUtils.readDnaQLazy(file);
+            int indexCutRead = 0;
+            for (DnaQ qual : quals) {
+                String read = qual.toString();
+                int len = read.length();
+                int kmersInRead = len - k + 1;
+                int kmersFiltration = kmersInRead * percentFiltration / 100;
+                kmersFiltration = Math.max(1, kmersFiltration);
+                int numberCoverageKmers = 0;
+                for (int i = 0; i < len - k; i++) {
+                    if (calc.isContainedInSubgraph(read.substring(i, i + k))) {
+                        numberCoverageKmers++;
+                    }
+                    if (numberCoverageKmers >= kmersFiltration) {
+                        indexCutRead++;
+                        out.println("@" + indexCutRead + "\n" + read + "\n+");
+                        for (int j = 0; j < len; j++) {
+                            out.print(qualF.getPhredChar(qual.phredAt(j)));
+                        }
+                        out.println();
+                        break;
+                    }
+                }
+            }
+            out.close();
+        } catch (IOException e) {
+            throw new ExecutionFailedException("Failed to read from file " + file.getPath());
+        }
+
+    }
+
+    @Override
+    public void run() {
+        try {
+            reader();
+        } catch (ExecutionFailedException e) {
+            logger.info(e.getMessage());
+        }
+    }
+
+}
+
+//--k 3 --reads reads.fastq --seq genes.fasta --maxkmers 100 --coverage 0 -o newfile

src/io/writers/GFAWriter.java

@@ -59,11 +59,11 @@ private void printEdge(PrintWriter out, SingleNode first, SingleNode second) {
         out.print("L\t");
         out.print((Math.min(first.rc.id, first.id) + 1) + (first.isGeneNode ? GENE_LABEL_SUFFIX : ""));
         out.print('\t');
-        out.print((first.id < first.rc.id ? "+" : "-"));
+        out.print((first.id < first.rc.id ? "-" : "+"));
         out.print('\t');
         out.print((Math.min(second.rc.id, second.id) + 1) + (second.isGeneNode ? GENE_LABEL_SUFFIX : ""));
         out.print('\t');
-        out.print((second.id > second.rc.id ? "+" : "-"));
+        out.print((second.id > second.rc.id ? "-" : "+"));
         out.print('\t');
         out.println((k - 1) + "M");
     }