#!/usr/bin/env perl

# Purpose: Substitute any instances of J, O or U with I, K and C, respectively.
#    This is to allow amino acid sequences containing these letters to be
#    accepted by WU BLAST 2006 and earlier.

# Input: FASTA format amino acid sequence(s)
# Standard Output: FASTA format, with letters substituted

# Installation: save the text of this file on your computer in a file named
# "aasubst.pl", then make the file executable with the shell command
# "chmod a+x aasubst.pl"

# Note: if this script fails to run, the problem can likely be fixed by replacing
# everything to the right of the exclamation mark on line 1 with the full path
# and filename of the perl interpreter on your computer.

# Example usage with xdformat:
# gunzip < nr.gz | aasubst.pl | xdformat -p -o nr -- -
#
# Example usage with blastp:
# aasubst.pl jousequence.aa | blastp nr - filter=seg echofilter
#     OR
# aasubst.pl < jousequence.aa | blastp nr - filter=seg echofilter

# Author: W. Gish
# Date: 2008-02-05

while (<>) {
	if (!/^>/) {
		tr/jJoOuU/iIkKcC/;
	}
	print;
}
exit 0;