#!/usr/bin/env perl
# Annotate proteins with hydrophobic amino acids, members of {VWFAMILY}.
#

# First, read in all lines of the protein sequence.
while (<>)
{
    chomp;
    $aa = $aa . $_;
}

# Translate all nucleotides to upper case
$aa =~ tr/a-z/A-Z/;

# Remove non-protein alphabet letters
#  s => substitute
#  g => globally; across the string
$aa =~ s/[^A-Z]//g;

# Change all the VWFAMILY letters to minus signs; hydrophobic
$aa =~ s/[VWFAMILY]/-/g;

# Change all remaining letters to positive signs; not hydrophobic
$aa =~ s/[A-Z]/+/g;

# Here, match $aa against 4-10 +'s, then combinations of between 17 and 33 -'s
while ($aa =~ /([\+\-]+?)((\+{1,2}\-+)+)/g) {
    $precursor = $1;
    $seq = $2;
    if (length($seq) <10 || length($seq) > 33) { next; }
    $scorestring = $seq;
    $scorestring =~ s/\-//g;
    $score =  (length($scorestring)/length($seq));
    if ($score > 0.43) { next; }

    $p = (pos $aa) - (length($seq));
    if (length($seq) == 0) { print "Woops!"; }
    print  length($seq),": ","<" . $precursor . ">", $p ,"[" . $seq . "]",($p+length($seq))," score: ",$score,"\n";

}