updates for genomespan ctor and fastasubseq.

compgen-io · Apr 17, 2015 · 58a1676 · 58a1676
1 parent a04d003
commit 58a1676
Show file tree

Hide file tree

Showing 6 changed files with 36 additions and 35 deletions.
diff --git a/lib/compgen-cmdline-0.4.1.jar b/lib/compgen-cmdline-0.4.1.jar
diff --git a/lib/compgen-common-0.1.jar b/lib/compgen-common-0.1.jar
diff --git a/src/java/io/compgen/ngsutils/NGSUtils.java b/src/java/io/compgen/ngsutils/NGSUtils.java
@@ -23,7 +23,7 @@
 import io.compgen.ngsutils.cli.bed.BedToBed3;
 import io.compgen.ngsutils.cli.bed.BedToBed6;
 import io.compgen.ngsutils.cli.bed.BedToFasta;
-import io.compgen.ngsutils.cli.fasta.FastaIndex;
+import io.compgen.ngsutils.cli.fasta.FastaSubseq;
 import io.compgen.ngsutils.cli.fasta.FastaTag;
 import io.compgen.ngsutils.cli.fastq.FastqFilterCli;
 import io.compgen.ngsutils.cli.fastq.FastqMerge;
@@ -72,7 +72,7 @@ public static void main(String[] args) {
                 .addCommand(FindEvents.class)
                 .addCommand(JunctionDiffCli.class)
                 .addCommand(FastaJunctions.class)
-                .addCommand(FastaIndex.class)
+                .addCommand(FastaSubseq.class)
                 .addCommand(PileupCli.class)
                 .addCommand(RepeatAnnotate.class)
                 .addCommand(GTFAnnotate.class)

diff --git a/src/java/io/compgen/ngsutils/annotation/GenomeSpan.java b/src/java/io/compgen/ngsutils/annotation/GenomeSpan.java
@@ -20,36 +20,25 @@ public class GenomeSpan implements Comparable<GenomeSpan> {
     final public int end;
     final public Strand strand;
 
-    public GenomeSpan(String ref, int start, int end, Strand strand){
+    public GenomeSpan(String ref, int start, int end, Strand strand) {
         super();
         this.ref = ref;
         this.start = start;
         this.end = end;
         this.strand = strand;
     }
 
+    public GenomeSpan(String ref, int start){
+        this(ref, start, Strand.NONE);
+    }
+
     public GenomeSpan(String ref, int start, int end){
-        super();
-        this.ref = ref;
-        this.start = start;
-        this.end = end;
-        this.strand = Strand.NONE;
+        this(ref, start, end, Strand.NONE);
     }
 
     public GenomeSpan(String ref, int start, Strand strand){
-        super();
-        this.ref = ref;
-        this.start = start;
-        this.end = start;
-        this.strand = strand;
-    }
-
-    public GenomeSpan(String ref, int start){
-        super();
-        this.ref = ref;
-        this.start = start;
-        this.end = start+1; // TODO: Should this be start + 1??? Check how this plays with comparisons.
-        this.strand = Strand.NONE;
+        // TODO: Should the end be start + 1??? Check how this plays with comparisons.
+        this(ref, start, start+1, Strand.NONE);
     }
 
     public GenomeSpan clone() {
@@ -97,6 +86,9 @@ public static GenomeSpan parse(String str, Strand strand, boolean zero) {
             return null;
         } else if (startend.indexOf('-') == -1) {
             int pos = Integer.parseInt(startend);
+            if (!zero) {
+                pos = pos - 1; // if this isn't zero based INPUT, then adjust the start position
+            }
             return new GenomeSpan(ref, pos, strand);
         } else {
             int start = Integer.parseInt(startend.substring(0, startend.indexOf('-')));
@@ -162,19 +154,23 @@ protected boolean contains(String qref, int qstart, int qend, Strand qstrand, bo
         }
         return false;
     }
-    
+
     @Override                                                                                                                                                                                                                                                     
     public String toString() {
+        return toString(false);
+    }
+
+    public String toString(boolean oneBasedStart) {
         if (start == end) {
             if (strand==Strand.NONE) {
-                return ref+":"+Integer.toString(start);
+                return ref+":"+Integer.toString(start + (oneBasedStart ? 1: 0));
             }
-            return ref+strand+":"+Integer.toString(start);
+            return ref+strand+":"+Integer.toString(start + (oneBasedStart ? 1: 0));
         }
         if (strand==Strand.NONE) {
-            return ref+":"+Integer.toString(start)+"-"+Integer.toString(end);
+            return ref+":"+Integer.toString(start + (oneBasedStart ? 1: 0))+"-"+Integer.toString(end);
         }
-        return ref+strand+":"+Integer.toString(start)+"-"+Integer.toString(end);
+        return ref+strand+":"+Integer.toString(start + (oneBasedStart ? 1: 0))+"-"+Integer.toString(end);
     }
 
     @Override                                                                                                                                                                                                                                                     
@@ -285,4 +281,5 @@ public GenomeSpan getStartPos() {
     public GenomeSpan getEndPos() {
         return new GenomeSpan(ref, end-1, end, strand);
     }
+
 }
diff --git a/...ompgen/ngsutils/cli/fasta/FastaIndex.java → ...mpgen/ngsutils/cli/fasta/FastaSubseq.java b/...ompgen/ngsutils/cli/fasta/FastaIndex.java → ...mpgen/ngsutils/cli/fasta/FastaSubseq.java
@@ -12,8 +12,8 @@
 import java.io.IOException;
 import java.util.List;
 
-@Command(name="faidx", desc="Extract subsequences from an indexed FASTA file", category="fasta")
-public class FastaIndex extends AbstractOutputCommand {
+@Command(name="fasta-subseq", desc="Extract subsequences from a FASTA file (optionally, indexed)", category="fasta", doc="Note: the start-position in the region should be 1-based")
+public class FastaSubseq extends AbstractOutputCommand {
 
     private String filename = null;
     private GenomeSpan region = null;
@@ -31,6 +31,9 @@ public void setArgs(List<String> args) throws CommandArgumentException {
         }
         filename = args.get(0);
         region = GenomeSpan.parse(args.get(1));
+        if (region.start < 0) {
+            throw new CommandArgumentException("Invalid region!");
+        }
     }
 
     @Exec
@@ -39,8 +42,8 @@ public void exec() throws IOException, CommandArgumentException {
             throw new CommandArgumentException("Missing/invalid arguments!");
         }
         FastaReader fasta = FastaReader.open(filename);
-        System.out.println(">"+region);
-        String seq = fasta.fetchSequence(region.ref, region.start-1, region.end);
+        System.out.println(">"+region.toString(true));
+        String seq = fasta.fetchSequence(region.ref, region.start, region.end);
         while (seq.length() > wrap) {
             System.out.println(seq.substring(0, wrap));
             seq = seq.substring(wrap);

diff --git a/src/java/io/compgen/ngsutils/fasta/BasicFastaReader.java b/src/java/io/compgen/ngsutils/fasta/BasicFastaReader.java
@@ -10,6 +10,7 @@
 public class BasicFastaReader extends FastaReader {
     protected final String filename;
     private boolean closed = false;
+    private StringLineReader reader = null;
 
     protected BasicFastaReader(String filename) {
         this.filename = filename;
@@ -29,10 +30,8 @@ public String fetchSequence(String ref, int start, int end) throws IOException {
             }
         }
 
-        while (it.hasNext()) {
-            // exhaust iterator to close open file
-            // should probably be something more efficient, but for random access, we should be indexed anyway.
-            it.next();
+        if (reader != null) {
+            reader.close();
         }
 
         return seq;
@@ -48,9 +47,11 @@ public Iterator<FastaRecord> iterator() throws IOException {
         if (closed) {
             throw new IOException("FastaReader closed");
         }
+
+        reader = new StringLineReader(filename);
+
         return new Iterator<FastaRecord> () {
 
-        StringLineReader reader = new StringLineReader(filename);
         Iterator<String> it = null;
 
         FastaRecord nextRecord = null;