Handle another type of shitty input MAF

maf2vcf.pl

@@ -134,6 +134,9 @@
     $al1 = $ref unless( defined $al1 );
     $al2 = $ref unless( defined $al2 );
 
+    # Handle a case when $al1 is a SNP we want to annotate, but $al2 is incorrectly "-"
+    ( $al1, $al2 ) = ( $al2, $al1 ) if( $al2 eq "-" );
+
     # To represent indels in VCF format, we need to fetch the preceding bp from a reference FASTA
     my ( $ref_len, $al1_len, $al2_len ) = map{( $_=~m/^(\?|-|0)+$/ ? 0 : length( $_ )) } ( $ref, $al1, $al2 );
     if( $ref_len == 0 or $al1_len == 0 or $al2_len == 0 ) {
@@ -163,8 +166,9 @@
     }
 
     # Set tumor and normal genotypes (FORMAT tag GT in VCF)
-    my $t_gt = join( "/", $al_idx{$al1}, $al_idx{$al2} );
-    my $n_gt = join( "/", $al_idx{$n_al1}, $al_idx{$n_al2} );
+    my ( $t_gt, $n_gt ) = ( "0/1", "0/0" ); # Set defaults
+    $t_gt = join( "/", $al_idx{$al2}, $al_idx{$al1} ) if( $al_idx{$al1} ne "0" );
+    $n_gt = join( "/", $al_idx{$n_al2}, $al_idx{$n_al1} ) if( $al_idx{$n_al1} ne "0" or $al_idx{$n_al2} ne "0" );
 
     # Create the VCF's comma-delimited ALT field that must list all non-REF (variant) alleles
     my $alt = join( ",", @alleles[1..$#alleles] );

vcf2maf.pl

@@ -496,6 +496,11 @@ sub GetBiotypePriority {
     # Figure out the appropriate start/stop loci and variant type/allele to report in the MAF
     my $start = my $stop = my $var_type = "";
     my ( $ref_length, $var_length ) = ( length( $ref ), length( $var ));
+    # Remove any prefixed reference bps from all alleles, using "-" for simple indels
+    while( substr( $ref, 0, 1 ) eq substr( $var, 0, 1 )) {
+        ( $ref, $var, @alleles ) = map{$_ = substr( $_, 1 ); ( $_ ? $_ : "-" )} ( $ref, $var, @alleles );
+        --$ref_length; --$var_length; ++$pos;
+    }
     # Handle SNPs, DNPs, TNPs, or anything larger (ONP)
     if( $ref_length == $var_length ) {
         ( $start, $stop ) = ( $pos, $pos + $var_length - 1 );
@@ -504,11 +509,6 @@ sub GetBiotypePriority {
     }
     # Handle all indels, including those complex ones which contain substitutions
     elsif( $ref_length != $var_length ) {
-        # Remove any prefixed reference bps from all alleles, using "-" for simple indels
-        while( substr( $ref, 0, 1 ) eq substr( $var, 0, 1 )) {
-            ( $ref, $var, @alleles ) = map{$_ = substr( $_, 1 ); ( $_ ? $_ : "-" )} ( $ref, $var, @alleles );
-            --$ref_length; --$var_length; ++$pos;
-        }
         if( $ref_length < $var_length ) { # Handle insertions, and the special case for complex ones
             ( $start, $stop ) = ( $pos - 1, ( $ref eq "-" ? $pos : $pos + $ref_length - 1 ));
             $var_type = "INS";