From 47c4a18a15d5f93d4f3622615b3368448a74127d Mon Sep 17 00:00:00 2001
From: Cyriac Kandoth <ckandoth@gmail.com>
Date: Thu, 23 Jul 2020 11:37:10 -0700
Subject: [PATCH] Fix maf2maf by using vcf2maf with inhibit-vep arg

---
 maf2maf.pl | 21 ++++++++++-----------
 1 file changed, 10 insertions(+), 11 deletions(-)

diff --git a/maf2maf.pl b/maf2maf.pl
index d9fc0d5..9875c65 100644
--- a/maf2maf.pl
+++ b/maf2maf.pl
@@ -109,10 +109,10 @@
     "--nrm-vad-col $nrm_vad_col --per-tn-vcfs";
 system( $maf2vcf_cmd ) == 0 or die "\nERROR: Failed to run maf2vcf! Command: $maf2vcf_cmd\n";
 
-my $vcf_file = "$tmp_dir/" . substr( $input_maf, rindex( $input_maf, '/' ) + 1 );
-$vcf_file =~ s/(\.)?(maf|tsv|txt)?$/.vcf/;
-my $vep_anno = $vcf_file;
-$vep_anno =~ s/\.vcf$/.vep.vcf/;
+my $tmp_basename = "$tmp_dir/" . substr( $input_maf, rindex( $input_maf, '/' ) + 1 );
+$tmp_basename =~ s/(\.)?(maf|tsv|txt)?$//;
+my $vcf_file = $tmp_basename . ".vcf";
+my $vep_anno = $tmp_basename . ".vep.vcf";
 
 # Skip running VEP if a VEP-annotated VCF already exists
 if( -s $vep_anno ) {
@@ -146,8 +146,7 @@
 }
 
 # Load the tumor-normal pairs from the TSV created by maf2vcf
-my $tsv_file = $vcf_file;
-$tsv_file =~ s/(\.vcf)?$/.pairs.tsv/;
+my $tsv_file = $tmp_basename . ".pairs.tsv";
 
 # Store the VEP annotated VCF header so we can duplicate it for per-TN VCFs
 my $vep_vcf_header = `grep ^## $vep_anno`;
@@ -220,13 +219,13 @@
     $tn_vep_fh->close;
 }
 
-# For each VCF generated by maf2vcf above, contruct a vcf2maf command and run it
-my @vcfs = grep{ !m/.vep.vcf$/ and !m/$vcf_file/ } glob( "$tmp_dir/*.vcf" ); # Avoid reannotating annotated VCFs
+# For each annotated VCF generated above, contruct a vcf2maf command and run it
+my @vcfs = grep{ !m/$tmp_basename/ } glob( "$tmp_dir/*.vep.vcf" );
 foreach my $tn_vcf ( @vcfs ) {
-    my ( $tumor_id, $normal_id ) = $tn_vcf=~m/^.*\/(.*)_vs_(.*)\.vcf/;
+    my ( $tumor_id, $normal_id ) = $tn_vcf=~m/^.*\/(.*)_vs_(.*)\.vep.vcf/;
     my $tn_maf = $tn_vcf;
-    $tn_maf =~ s/.vcf$/.vep.maf/;
-    my $vcf2maf_cmd = "$perl_bin $vcf2maf_path --input-vcf $tn_vcf --output-maf $tn_maf" .
+    $tn_maf =~ s/.vep.vcf$/.vep.maf/;
+    my $vcf2maf_cmd = "$perl_bin $vcf2maf_path --input-vcf $tn_vcf --output-maf $tn_maf --inhibit-vep" .
         " --tumor-id $tumor_id --normal-id $normal_id --vep-path $vep_path --vep-data $vep_data" .
         " --vep-forks $vep_forks --ref-fasta $ref_fasta --ncbi-build $ncbi_build --species $species" .
         " --filter-vcf $filter_vcf --max-filter-ac $max_filter_ac";