X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=misc%2Fsam2vcf.pl;h=afaf91e1dcf60894f195bed3580b6712dc0457df;hb=0b3418cf166ce4a58cedf0d9a2df5ec3dd4cc5fa;hp=c2644f53d1210669c04bbbdb02559b0b966259cc;hpb=8443a8179d04a83464fe514e8b9b40e7f456b0fe;p=samtools.git

diff --git a/misc/sam2vcf.pl b/misc/sam2vcf.pl
index c2644f5..afaf91e 100755
--- a/misc/sam2vcf.pl
+++ b/misc/sam2vcf.pl
@@ -3,7 +3,7 @@
 # VCF specs: http://www.1000genomes.org/wiki/doku.php?id=1000_genomes:analysis:vcf3.3
 # 
 # Contact: pd3@sanger
-# Version: 2009-12-08
+# Version: 2010-04-23
 
 use strict;
 use warnings;
@@ -23,10 +23,12 @@ sub error
     die
         "Usage: sam2vcf.pl [OPTIONS] < in.pileup > out.vcf\n",
         "Options:\n",
-        "   -r, --refseq <file.fa>           The reference sequence, required when indels are present.\n",
-        "   -s, --snps-only                  Ignore indels.\n",
-        "   -t, --column-title <string>      The column title.\n",
-        "   -h, -?, --help                   This help message.\n",
+        "   -h, -?, --help                  This help message.\n",
+        "   -i, --indels-only               Ignore SNPs.\n",
+        "   -r, --refseq <file.fa>          The reference sequence, required when indels are present.\n",
+        "   -R, --keep-ref                  Print reference alleles as well.\n",
+        "   -s, --snps-only                 Ignore indels.\n",
+        "   -t, --column-title <string>     The column title.\n",
         "\n";
 }
 
@@ -40,9 +42,11 @@ sub parse_params
 
     while (my $arg=shift(@ARGV))
     {
+        if ( $arg eq '-R' || $arg eq '--keep-ref' ) { $opts{keep_ref}=1; next; }
         if ( $arg eq '-r' || $arg eq '--refseq' ) { $opts{refseq}=shift(@ARGV); next; }
         if ( $arg eq '-t' || $arg eq '--column-title' ) { $opts{title}=shift(@ARGV); next; }
         if ( $arg eq '-s' || $arg eq '--snps-only' ) { $opts{snps_only}=1; next; }
+        if ( $arg eq '-i' || $arg eq '--indels-only' ) { $opts{indels_only}=1; next; }
         if ( $arg eq '-?' || $arg eq '-h' || $arg eq '--help' ) { error(); }
 
         error("Unknown parameter \"$arg\". Run -h for help.\n");
@@ -103,6 +107,8 @@ sub do_pileup_to_vcf
     my ($prev_chr,$prev_pos,$prev_ref);
     my $refseq;
     my $ignore_indels = $$opts{snps_only} ? 1 : 0;
+    my $ignore_snps   = $$opts{indels_only} ? 1 : 0;
+    my $keep_ref      = $$opts{keep_ref} ? 1 : 0;
     my $title = exists($$opts{title}) ? $$opts{title} : 'data';
 
     print $fh_out 
@@ -117,28 +123,43 @@ sub do_pileup_to_vcf
     while (my $line=<$fh_in>)
     {
         chomp($line);
-        my ($chr,$pos,$ref,$cons,$cons_qual,$snp_qual,$rms_qual,$depth,@items) = split(/\t/,$line);
+        my (@items) = split(/\t/,$line);
+        if ( scalar @items<8 ) 
+        { 
+            error("\nToo few columns, does not look like output of 'samtools pileup -c': $line\n"); 
+        }
+        my ($chr,$pos,$ref,$cons,$cons_qual,$snp_qual,$rms_qual,$depth,$a1,$a2) = @items;
+        $ref  = uc($ref);
+        $cons = uc($cons);
 
         my ($alt,$gt);
         if ( $ref eq '*' )
         {
             # An indel is involved.
-            if ( $ignore_indels ) { next; }
+            if ( $ignore_indels )
+            { 
+                $prev_ref = $ref;
+                $prev_pos = $pos;
+                $prev_chr = $chr;
+                next; 
+            }
 
-            if ($chr ne $prev_chr || $pos ne $prev_pos) 
+            if (!defined $prev_chr || $chr ne $prev_chr || $pos ne $prev_pos) 
             {
                 if ( !$$opts{refseq} ) { error("Cannot do indels without the reference.\n"); }
                 if ( !$refseq ) { $refseq = Fasta->new(file=>$$opts{refseq}); }
                 $ref = $refseq->get_base($chr,$pos);
+                $ref = uc($ref);
             }
             else { $ref = $prev_ref; }
 
-            # One of the alleles can be a reference and it can come in arbitrary order
+            # One of the alleles can be a reference and it can come in arbitrary order. In some
+            #   cases */* can be encountered. In such a case, look in the additional columns.
             my ($al1,$al2) = split(m{/},$cons);
+            if ( $al1 eq $al2 && $al1 eq '*' ) { $al1=$a1; $al2=$a2; }
             my $alt1 = parse_indel($al1);
             my $alt2 = parse_indel($al2);
             if ( !$alt1 && !$alt2 ) { error("FIXME: could not parse indel:\n", $line); }
-            if ( $alt1 && $alt2 && $alt1 eq $alt2 ) { $alt2=''; }
             if ( !$alt1 ) 
             { 
                 $alt=$alt2; 
@@ -149,7 +170,12 @@ sub do_pileup_to_vcf
                 $alt=$alt1; 
                 $gt='0/1'; 
             }
-            else 
+            elsif ( $alt1 eq $alt2 )
+            { 
+                $alt="$alt1"; 
+                $gt='1/1'; 
+            }
+            else
             { 
                 $alt="$alt1,$alt2"; 
                 $gt='1/2'; 
@@ -157,6 +183,14 @@ sub do_pileup_to_vcf
         }
         else
         {
+            if ( $ignore_snps || (!$keep_ref && $ref eq $cons) ) 
+            { 
+                $prev_ref = $ref;
+                $prev_pos = $pos;
+                $prev_chr = $chr;
+                next; 
+            }
+
             # SNP
             ($alt,$gt) = iupac_to_gtype($ref,$cons);
         }