#!/home/aaghili/perl5
#### Script to scrape genes for each pathway
# Ardesher Aghili's first big boy script :^)
use warnings;
use strict;
#### UPDATE 7/21/17 - SCRIPT WILL PRINT AN ARRAY OF VALUES FROM THE FIRST COLUMN OF A CSV!
#### UPDATE 7/24/17 - SCRIPT WILL WRITE TO FILE, BUT HAS SPACES INSTEAD OF COMMAS!
my $file0 = $ARGV[0] or die "Must target a CSV file\n";
my @pathways = "";
open(my $data0, '<', $file0) or die "Could not open '$file0' $!\n";
while (my $line0 = <$data0>)  {
    # pathway name extraction
    chomp $line0;

    my @fields0 = split "," , $line0;
    push @pathways, $fields0[1];
}

my $file1 = $ARGV[1] or die "No gene list selected\n";
open(my $data1, '<', $file1) or die "Could not open gene list '$file1' $!\n";
while (my $line1 = <$data1>)   {
    chomp $line1;

    my @fields1 = split "," , $line1;
    foreach $pathways[i] (@pathways) {
        #match pathway name with first elelment of line
        if ($pathways[i] = $fields[0]) {
            #write array for pathway
            my @pathways[i] = "$pathways[i]";
            while $fields[i] != ""
                push @pathways[i], $fields[1];
        }
    }
}

# Writing genes to CSV
open (my $output, '+<', 'nominal_pathways.csv');
print $output "@pathways"; #### THESE ARE ONLY THE PATHWAY NAMES
close $output;
print "Created nominal_pathways.csv \n";