annotate combine_hgvs_tables @ 0:baf1543e8ae1 default tip

initial commit

#!/usr/bin/env perl

use strict;
use warnings;

if(@ARGV == 1 and $ARGV[0] eq "-v"){
  print "Version 1.0\n";
  exit;
}

my $quiet = 0;
if(@ARGV and $ARGV[0] =~ /^-q(?:uiet)?$/){
  $quiet = 1;
  shift @ARGV;
}

@ARGV >= 6 and not @ARGV%2 or die "Usage: $0 [-q(uiet)] <true|false> <output file> <method_name1> <hgvs_file_1.txt> <method_name2> <hgvs_file_2.txt> ...\n",
                                  "Where true or false determines if the genotypes calls should all be reported (if false, they are collapsed to the unique set of values observed)\n";

my %zygosity;
my %quality;
my %var_count;
my %tot_count;
my %methods;
my %data;
my $chr_prefix_exists = 0; # do any of the tools report chr1, if so we need to ensure 1 get reported as chr1 later for consistency across tools

my $collapse = $ARGV[0] =~ /^true$/i;

open(OUT, ">$ARGV[1]")
  or die "Cannot open $ARGV[1] for writing: $!\n";

my @headers;
my ($chr_column, $zygosity_column, $pvalue_column, $alt_cnt_column, $tot_cnt_column, $transcript_column, $cdna_hgvs_column, $pos_column, $to_column, $srcs_column);
for(my $i=2; $i<$#ARGV; $i+=2){
  my $method_name = $ARGV[$i];
  my $infile = $ARGV[$i+1];
  print STDERR "Processing $infile...\n" unless $quiet;
  open(IN, $infile)
    or die "Cannot open $infile for reading: $!\n";
  my $header = <IN>; # header
  chomp $header;
  if($i == 2){
    @headers = split /\t/, $header;
    for(my $j = 0; $j <= $#headers; $j++){
      if($headers[$j] eq "Chr"){
        $chr_column = $j;
      }
      elsif($headers[$j] eq "Zygosity"){
        $zygosity_column = $j;
      }
      elsif($headers[$j] eq "P-value"){
        $pvalue_column = $j;
      }
      elsif($headers[$j] eq "Variant Reads"){
        $alt_cnt_column = $j;
      }
      elsif($headers[$j] eq "Total Reads"){
        $tot_cnt_column = $j;
      }
      elsif($headers[$j] eq "Selected transcript"){
        $transcript_column = $j;
      }
      elsif($headers[$j] eq "Transcript HGVS"){
        $cdna_hgvs_column = $j;
      }
      elsif($headers[$j] eq "DNA From"){
        $pos_column = $j;
      }
      elsif($headers[$j] eq "DNA To"){
        $to_column = $j;
      }
      elsif($headers[$j] eq "Sources"){
        $srcs_column = $j;
      }
    }
    if(defined $srcs_column){
      # print header as-is
      print OUT "$header\n";
    }
    else{
      # print header from the first file, with extra column for methods at the end
      push @headers, "Sources";
      $srcs_column = $#headers;
      print OUT join("\t", @headers), "\n";
    }
  }
  else{
    # Make sure the columns are the same in the various files
    my @latest_headers = split /\t/, $header;
    for(my $j = 0; $j <= $#latest_headers && $j <= $#headers; $j++){
      if($latest_headers[$j] ne $headers[$j]){
        die "Header column ", $j+1, "($latest_headers[$j]) in $ARGV[$i] is different from the equivalent column label ($headers[$j]) in $ARGV[2]. Aborting, cannot merge the files.\n";
      }
    }
  }
  while(<IN>){
    chomp;
    my @F = split /\t/, $_, -1; # -1 to keep trailing blank fields
    if(not $chr_prefix_exists and $F[$chr_column] =~ /^chr/){
       $chr_prefix_exists = 1;
    }
    $F[$chr_column] =~ s/^chr//; # for consistency
    my $key = "$F[$transcript_column]:$F[$cdna_hgvs_column]"; # transcript_id:cdna_hgvs is shared id for common variants amongst files
    # record disagreement on zygosity if any
    if(not defined $zygosity{$key}){
      $zygosity{$key} = [];
      $quality{$key} = [];
      $var_count{$key} = [];
      $tot_count{$key} = [];
      $data{$key} = \@F;
      #print STDERR "Missing fields (only have ", scalar(@F), " for input '$_'\n" if $#F < $#headers;
    }
    push @{$zygosity{$key}}, split(/; /,$F[$zygosity_column]);
    push @{$quality{$key}}, $F[$pvalue_column];
    push @{$var_count{$key}}, $F[$alt_cnt_column];
    push @{$tot_count{$key}}, $F[$tot_cnt_column];
    push @{$methods{$key}}, $method_name;
  }
  close(IN);
}

if($chr_prefix_exists){
  for my $key (keys %data){
    # assuming there is a mix of chr1 and 1, always add the chr for consistency
    $data{$key}->[$chr_column] = "chr".$data{$key}->[$chr_column] unless $data{$key}->[$chr_column] =~ /^chr/;
  }
}

# Sort by chr, then pos, then transcript_id
for my $key (sort {$data{$a}->[$chr_column] cmp $data{$b}->[$chr_column] or 
                   $data{$a}->[$pos_column] <=> $data{$b}->[$pos_column] or 
                   $data{$a}->[$to_column] <=> $data{$b}->[$to_column] or 
                   $data{$a}->[$transcript_column] cmp $data{$b}->[$transcript_column]} keys %data){
  # Before printing a line, merge the zygosity, variant quality, read count data if requested
  my %seen;
  if($collapse){
    my @zygosities = grep {$_ ne "NA" and not $seen{$_}++} @{$zygosity{$key}};
    if(@zygosities == 0){
      # do nothing, existing NA in 7th column will be a fine answer as they are all like tha
    }
    elsif(@zygosities == 1){
      # agreement by all methods, just print the one not NA
      $data{$key}->[$zygosity_column] = $zygosities[0] if $data{$key}->[$zygosity_column] eq "NA";
    }
    else{
      $data{$key}->[$zygosity_column] = join("/", sort keys %seen); # list all unique values that occur
    }
  }
  else{
    $data{$key}->[$zygosity_column] = join("; ", @{$zygosity{$key}});
  }

  if($collapse){
    undef %seen;
    my @qualities = grep {$_ ne "NA" and not $seen{$_}++} @{$quality{$key}};
    if(@qualities == 0){
      # do nothing, existing NA in 8th column will be a fine answer as they are all like that
    }
    elsif(@qualities == 1){
      # agreement by all methods, just print the one not NA
      $data{$key}->[$pvalue_column] = $qualities[0] if $data{$key}->[8] eq "NA";
    }
    else{
      # calculate the median for the collapsed value
      my @sorted_qualities = sort {$a <=> $b} grep({$_ ne "NA"} @{$quality{$key}});
      my $median_quality = $sorted_qualities[int($#sorted_qualities/2)]; # rounds down (actually better score as this is a p-value)
      $data{$key}->[$pvalue_column] = $median_quality;
    }
  }
  else{
    $data{$key}->[$pvalue_column] = join("; ", @{$quality{$key}});
  }

  if($collapse){
    undef %seen;
    my @var_counts = grep {$_ ne "NA" and not $seen{$_}++} @{$var_count{$key}};
    undef %seen;
    my @tot_counts = grep {$_ ne "NA" and not $seen{$_}++} @{$tot_count{$key}};
    if(@var_counts == 0 and @tot_counts == 0){
      # do nothing, existing NAs okay
    }
    elsif(@var_counts == 1 and @tot_counts == 1){
      # agreement by all methods, just print the one in %data unless it's NA
      $data{$key}->[$alt_cnt_column] = $var_counts[0] if $data{$key}->[$alt_cnt_column] eq "NA";
      $data{$key}->[$tot_cnt_column] = $tot_counts[0] if $data{$key}->[$tot_cnt_column] eq "NA";
    }
    else{
      # calculate the median for the collapsed value
      my @sorted_var_counts = sort {$a <=> $b} grep({$_ ne "NA"} @{$var_count{$key}});
      my $pivot = $#sorted_var_counts/2;
      my $median_var_count = $pivot == int($pivot) ? $sorted_var_counts[$pivot] : # arithmetic mean
                                                     int(($sorted_var_counts[int($pivot)]+$sorted_var_counts[int($pivot)+1])/2);
      $data{$key}->[$alt_cnt_column] = $median_var_count;
      my @sorted_tot_counts = sort {$a <=> $b} grep({$_ ne "NA"} @{$tot_count{$key}});
      $pivot = $#sorted_tot_counts/2;
      my $median_tot_count = $pivot == int($pivot) ? $sorted_tot_counts[$pivot] : # arithmetic mean
                                                     int(($sorted_tot_counts[int($pivot)]+$sorted_tot_counts[int($pivot)+1])/2);
      $data{$key}->[$tot_cnt_column] = $median_tot_count;
    }
  }
  else{
    # list the raw numbers
    $data{$key}->[$alt_cnt_column] = join("; ", @{$var_count{$key}});
    $data{$key}->[$tot_cnt_column] = join("; ", @{$tot_count{$key}});
  }

  # to facilitate multiple rounds of combining, slash the extra column from the last round if it exists
  $data{$key}->[$srcs_column] = join("; ", @{$methods{$key}});
  for(my $i = 0; $i <= $#{$data{$key}}; $i++){
    $data{$key}->[$i] = "" if not defined $data{$key}->[$i];
  }
  print OUT join("\t", @{$data{$key}}), "\n";
}
close(OUT);