#!/usr/local/bin/perl -w

# msf2alpro.pl

# Usage:  msf2alpro.pl   file.msf  >  protseq

# convert MSF multiple alignment format to
# alpro format to generate sequence logos.
# Mike Sauder, msauder@stromix.com   11-29-00

# Replacing > with * is not necessary in newest version of 'alpro'
# Append . to end of sequences; Replace . with - and remove spaces.

%seqs=(); @line=(); $ok=0; 
$width=60;

if (!defined($ARGV[0]) || $ARGV[0] eq "-h" || $ARGV[0] eq "-help") { printhelp() }
$filename=$ARGV[0];

while(<>) {
    chomp;
    if (/^\/\//) { $ok=1; next }
    if (/^     /) { next }
    if ($_ ne "" && $ok==1) {
        @line=split;
        $name=$line[0];
        for ($i=1; $i<6; $i++) {
            if (defined($line[$i])) { $seqs{$name} .= $line[$i] }
        }
    }
}
if ($ok==0) { 
    print STDERR "* Not an MSF alignment file; no // line.\n";
    print STDERR "Please specify an MSF alignment file on the command line.\n";
    die "\n";
}
else { print "* ${filename}; alpro format for use with Sequence Logos (seqlogo).\n" }


foreach $name (sort keys %seqs) {
    print ">$name \n";
    $seqs{$name}=~s/\./\-/g;  # Convert . to -
    $seqs{$name}.=".";
    printseq($seqs{$name});
}


# Format sequence at $width columns
sub printseq {
    $seq = $_[0];
    $length=length($seq);
    $ncol=int($length/$width);
    for ($i=0; $i<=$ncol; $i++) {
        print substr($seq,$i*$width,$width),"\n";
    }
}


sub printhelp {
    print<<'EOF';

NAME
      msf2alpro.pl - Convert MSF alignment format to alpro format

USAGE
      msf2alpro.pl  file.msf  >  protseq

DESCRIPTION
      msf2alpro.pl converts a MSF format multiple alignment file to
      alpro format for generating Sequence Logos based on the 'alpro' 
      and 'makelogo' programs by Tom Schneider.

EOF
    die "\n";
}


=pod

# protseq.demo - sample alpro format:

* protseq.globin: aligned globin sequences
>Name1 description
---------------DLVSGFWGKV--DA---HKIGGEALARLLVVYPWTQRYFTTFGNL
GSADAIC-----HNA---KVLAHG-EKVLAAIGEGLKHPENLKAHY--AKL-SEYHSNK-
---LHVDPANFRLLGNVFITVLARHF-QH-EFTPELQ-.
>Name2 description
---------LTAAEKTQVTNLWGKV--NV---KELGGEALSRLLVVYPWTRRFFEHFGDL
STAEAVL-----HNA---KVLAHG-EKVLTSFGEGLKHLDNLKGTF--ADL-SELHCDK-
---LHVDPENFRLLGNVLVIVLARHF-GK-EFTPDVQ-.
>Name3 description
---------LTAEEKAAVTGFWGKV--KV---DEVGAEALGRLLVVYPWTQRFFEHFGDL
SSADAVM-----NNA---KVKAHG-KKVLDSFSNGMKHLDDLKGTF--AQL-SELHCDK-
---LHVDPENFKLLGNVLVVVLARHH-GS-EFTPLLQ-A.

=cut