リン酸化部位とかUniProtのフラットファイルからとる

gistfile1.pl

#!/usr/bin/env perl

use strict;
use warnings;

use Data::Dumper;

open my $fh, '<', $ARGV[0] or die;
local $/ = undef;

my $data = <$fh>;
my @flat = split( m{//\n}, $data );

my $uni = {};
for my $line (@flat) {
    
    #Parse ID line
    $line  =~ m/ID\s+(.+\_YEAST)\s+Reviewed\;\s+\d+\w+/;
    my $ac = $1 if defined $line;
    
  THREONINE:
    while ( $line =~ /FT\s+MOD_RES\s+(\d+)\s+\d+\s+(Phosphothreonine)/g ) {
	my $pos     = $1;
	my $mod_res = $2;
	
	unless ( !$pos && !$mod_res ) {
	    push @{ $uni->{$ac}->{THREONINE} }, $pos;
	    
	    #Parse DE line
	    $line =~ m/DE\s+(.+)/;
	    $uni->{$ac}->{DE} = $1 if defined $1;
	}
	else {
	    next THREONINE;
	}
    }
    
  TYROSINE:
    while ( $line =~ /FT\s+MOD_RES\s+(\d+)\s+\d+\s+(Phosphotyrosine)/g ) {
	my $pos     = $1;
	my $mod_res = $2;
	
	unless ( !$pos && !$mod_res ) {
	    push @{ $uni->{$ac}->{TYROSINE} }, $pos;
	}
	else {
	    next TYROSINE;
	}
    }
    
  SERINE:
    while ( $line =~ /FT\s+MOD_RES\s+(\d+)\s+\d+\s+(Phosphoserine)/g ) {
	my $pos     = $1;
	my $mod_res = $2;
	
	unless ( !$pos && !$mod_res ) {
	    push @{ $uni->{$ac}->{SERINE} }, $pos;
	}
	else {
	    next SERINE;
	}
    }
}