comparison rapsodyn/listfiltering.pl @ 29:7b8646f46010 draft

Uploaded

28:fefe6d89dbf1

#!/usr/bin/perl
use strict;
use Getopt::Long;

my $inputfile;
my $headerfile;
my $nb_col=1;
my %header;

if ($#ARGV<0){
	print "\n";
	print "perl 021_ListFiltering.pl -input_file <file> -header_file <file> -nb_col <integer>[1]\n";
	exit(0);
}

GetOptions (
"input_file=s" => \$inputfile,
"header_file=s" => \$headerfile,
"nb_col=i" => \$nb_col
) or die("Error in command line arguments\n");

open(HF, $headerfile)  or die("Can't open $headerfile\n");
while (my $line=<HF>){
	chomp($line);
	my @fields = split(/\s+/,$line);
	my $ref="";
	my $compt=0;
	while ($compt<$nb_col){
		if ($ref){$ref.="\t";}
		$ref.=$fields[$compt];
		$compt++;
	}
	# my $ref = "$fields[0]\t$fields[1]";
	$header{$ref}=$line;
}
close (HF);


open(IF, $inputfile)  or die("Can't open $inputfile\n");
while (my $line=<IF>){
	my @fields = split(/\s+/,$line);
	my $ref="";
	my $compt=0;
	while ($compt<$nb_col){
		if ($ref){$ref.="\t";}
		$ref.=$fields[$compt];
		$compt++;
	}
	# my $ref = "$fields[0]\t$fields[1]";

	if ($header{$ref}){
		# print $line;
		# print $header{$ref},"\n";
	}
	else {
		print $line;
	}
	
}
close(IF);

29:7b8646f46010

#!/usr/bin/perl
# v1.0.1 added log, and two different type of filtering (common / specific)
use strict;
use Getopt::Long;

my $list1_file;
my $list2_file;
my $log_file;
my $NB_COL=1;
my $TYPE = "common";
my %header;
my $nb_list1 = 0;
my $nb_list2 = 0;
my $nb_common = 0;


GetOptions (
"list1_file=s" => \$list1_file,
"list2_file=s" => \$list2_file,
"log_file=s" => \$log_file,
"type=s" => \$TYPE,
"nb_col=i" => \$NB_COL
) or die("Error in command line arguments\n");

open(L2, $list2_file)  or die("Can't open $list2_file\n");
while (my $line=<L2>){
	$nb_list2++;
	chomp($line);
	my @fields = split(/\s+/,$line);
	my $ref="";
	my $compt=0;
	while ($compt<$NB_COL){
		if ($ref){$ref.="\t";}
		$ref.=$fields[$compt];
		$compt++;
	}
	$header{$ref}=$line;
}
close (L2);


open(L1, $list1_file)  or die("Can't open $list1_file\n");
while (my $line=<L1>){
	$nb_list1++;
	my @fields = split(/\s+/,$line);
	my $ref="";
	my $compt=0;
	while ($compt<$NB_COL){
		if ($ref){$ref.="\t";}
		$ref.=$fields[$compt];
		$compt++;
	}
	# my $ref = "$fields[0]\t$fields[1]";

	if ($header{$ref}){
		$nb_common++;
		if ($TYPE eq "common"){
			print $line;
		}
		elsif ($TYPE eq "specific") {
		}
		else {
		}
	}
	else {
		if ($TYPE eq "common"){
		}
		elsif ($TYPE eq "specific") {
			print $line;
		}
		else {
		}
	}
	
}
my $nb_list1_only = $nb_list1 - $nb_common;
my $nb_list2_only = $nb_list2 - $nb_common;

close(L1);
open (LF,">$log_file") or die("Can't open $log_file\n");
print LF "\n####\t List Filtering \n";
print LF "#List 1 :\t$nb_list1 ($nb_list1_only)\n";
print LF "#List 2 :\t$nb_list2 ($nb_list2_only)\n";
print LF "#Common :\t$nb_common\n";
close (LF);