root/bin/maker_functional_fasta

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

#-----------------------------------------------------------------------------
#----------------------------------- MAIN ------------------------------------
#-----------------------------------------------------------------------------

my $usage = "

Synopsis:

maker_functional_fasta uniprot-sprot.fasta uniprot-sprot.blast.out \
                       maker.transcripts.fasta maker.proteins.fasta ....

Description:

This script will take a uniprot-sprot fasta file, a wu-blast -mformat
2 blast output file from a blast of maker proteins against
uniprot-sprot proteins and add functional annotations to the maker
fasta header changing it from this:

>PYU1_T015316 transcript offset:0 AED:0.0476397966594044 \
QI:0|-1|0|1|-1|1|1|0|255

to this:

>PYU1_T015316 transcript name:\"Serine-aspartate repeat-containing protein I \
(Staphylococcus saprophyticus)\" offset:0 AED:0.0476397966594044             \
QI:0|-1|0|1|-1|1|1|0|255

";


my ($help);
my $opt_success = GetOptions('help'    => \$help,
                            );

die $usage if $help || ! $opt_success;

my ($uniprot_file, $blast_file, @fasta_files) = @ARGV;
die $usage unless $uniprot_file && $blast_file && @fasta_files;

my $blast = parse_blast($uniprot_file, $blast_file);

for my $file (@fasta_files) {

open (my $IN,  '<', $file) or die "Can't open $file for reading\n$!\n";

LINE:
while (<$IN>) {

    if ($_ !~ /^>/) {
        print $_;
        next LINE;
    }

    my ($id, $mol_type, @attrbs) = split /\s+/, $_;
    $id =~ s/^>//;

    my $func = $blast->{$id} || 'Name:"Protein of unknown function"';

    my $header = ">$id $mol_type ";
    $header .= "$func " if $func;
    $header .= join " ", @attrbs if @attrbs;

    print "$header\n";

    }
}
#-----------------------------------------------------------------------------
#-------------------------------- SUBROUTINES --------------------------------
#-----------------------------------------------------------------------------

sub parse_blast{

        my ($fasta_file, $blast_file) = @_;

        open(my $IN, '<', $fasta_file) or
          die "Can't open $fasta_file for reading\n$!\n";

        my %fasta;
        while (<$IN>) {

                next unless /^>/;

                my ($id, $desc, $org, $name);
                #>sp|P18560|1101L_ASFB7 Protein MGF 110-1L OS=African swine fever virus (strain Badajoz 1971 Vero-adapted) GN=BA71V-008 PE=3 SV=1
                if (/>(\S+)\s+(.*?)\s+OS=(.*?)\s+(GN=(.*?)\s+)?PE=/) {
                        $id   = $1;
                        $desc = $2;
                        $org  = $3;
                        $name = $5 || '';
                }
                else {
                        warn "Can't parse details from FASTA header: $_\n";
                }

                my $func = '';
                $func .= "Similar to $name: " if $name;
                if ($func =~ /Similar to/) {
                        $func .= "$desc "  if $desc;
                }
                else {
                        $func .= "Similar to $desc "  if $desc;
                }
                $func .= "($org)"  if $org;
                $func  = ('Name:"' . $func . '"') if $func;

                $fasta{$id} = $func;

                print '';
        }
        close $IN;

        open($IN, '<', $blast_file) or
          die "Can't open $blast_file for reading\n$!\n";

        my (%blast, %seen);
        while (<$IN>) {

                my ($qid, $sid, $e_val) = split;

                next if $seen{$qid};

                $seen{$qid}++;

                $blast{$qid} = $fasta{$sid} || '';
        }
        return \%blast;
}