From 735c1f73ea8da26613f905b6138a14ed0e343e75 Mon Sep 17 00:00:00 2001
From: Cyriac Kandoth <ckandoth@gmail.com>
Date: Thu, 29 Sep 2016 12:52:22 -0400
Subject: [PATCH] Fix ref matching for long indels

---
 maf2vcf.pl | 4 +++-
 vcf2maf.pl | 4 +++-
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/maf2vcf.pl b/maf2vcf.pl
index d01896d..b754f90 100644
--- a/maf2vcf.pl
+++ b/maf2vcf.pl
@@ -101,7 +101,9 @@
 push( @regions_split, [ splice( @regions, 0, 100000 ) ] ) while @regions;
 map{ my $loci = join( " ", @{$_} ); $lines .= `$samtools faidx $ref_fasta $loci` } @regions_split;
 foreach my $line ( grep( length, split( ">", $lines ))) {
-    my ( $locus, $bps ) = split( "\n", $line );
+    # Carefully split this FASTA entry, properly chomping newlines for long indels
+    my ( $locus, $bps ) = split( "\n", $line, 2 );
+    $bps =~ s/\r|\n//g;
     if( $bps ){
         $bps = uc( $bps );
         $flanking_bps{$locus} = $bps;
diff --git a/vcf2maf.pl b/vcf2maf.pl
index 768fd30..4756d35 100644
--- a/vcf2maf.pl
+++ b/vcf2maf.pl
@@ -248,7 +248,9 @@ sub GetBiotypePriority {
 push( @regions_split, [ splice( @regions, 0, 100000 ) ] ) while @regions;
 map{ my $loci = join( " ", @{$_} ); $lines .= `$samtools faidx $ref_fasta $loci` } @regions_split;
 foreach my $line ( grep( length, split( ">", $lines ))) {
-    my ( $locus, $bps ) = split( "\n", $line );
+    # Carefully split this FASTA entry, properly chomping newlines for long indels
+    my ( $locus, $bps ) = split( "\n", $line, 2 );
+    $bps =~ s/\r|\n//g;
     if( $bps ){
         $bps = uc( $bps );
         $flanking_bps{$locus} = $bps;