#!/usr/local/bin/gawk -f
#
#@ convert protein aminoacid sequence files from 
#@ oneletter to threeletter format and vice versa.
#
# (c) Eiso AB
#
# command line options:
# [-v] mode=3to1   # convert from three to one letter format
# [-v] mode=1to3   # convert from one to three letter format [ABCDE...]
# [-v] mode=1nto3   # convert from one to three letter format [A1/
#
# seq2seq.awk will try to determine the type of file from the first
# field on the first line it encounters. If the first word has either 
# one or 4 or more characters oneletter format will be assumed.
# 
# possible input:
#

BEGIN { 
    mode=""  # "3to1" or "1to3"
    progname="seq2seq.awk"
    if ( ARGV[1]~/^(h|help)$/ || help!="" ) { 
        print ARGV[0]
        # because of the which command, help is only available if 
        # this script is in the PATH
        system("gawk '/^#/{print}/^BEGIN/{exit}' `which "progname"`")
        help=1
        
        exit
    }
} 
#{ sub(/[\r]$/,"") }

mode=="" { # determine mode
    if ($0~/^[A-Z][0-9]+[\r]?$/) mode="1nto3"  # pasta sequence
    else if ($1~/^([A-Z][A-Z][A-Z0-9][A-Z0-9]?)[-+]?$/) mode="3to1"
    else if (length($1)==1) mode="1to3"
    else if (length($1)>1 && $1~/^[A-Z][0-9]+$/) mode="1to3"
    else if (length($1)>=4) mode="1to3"
    else mode="?"
}

# get sequence from sparky project file:
FILENAME ~ /[.]proj$/ { 
    if ($1!="sequence") next
    $0=$2
    #print "##??",$0
    mode="1to3"
    for (i=1;i<=length($0);i++) {
        aa=substr($f,i,1)
        if (aa~/^[A-Z]/) {
            print to_3_letter(aa),++r,(i==1?"# "ww:"")
        }
    }
    next
}

{   #print mode,$0
    sub(/#.*/,""); $1=$1; $0=$0
    if (mode=="3to1") { 
        for (f=1;f<=NF;f++) { ww=$0
            if ($f!~/^[0-9]+$/)
                printf "%s\n",to_1_letter($f)  (f==1?" # "ww:"")
        }
    } else if (mode=="1to3") {
        for (f=1;f<=NF;f++) { ww=$f
            for (i=1;i<=length($f);i++) {
                aa=substr($f,i,1)
                if (aa~/^[A-Z]/)
                    print to_3_letter(aa),(i==1?"# "ww:"")
            }
        }
    } else if (mode=="1nto3") {
        sub("[0-9]"," &")
        $0=$0
        if (NF==2) print to_3_letter($1),$2+0
    
    } else {
        print "#??",$0
    }
}

function to_3_letter(str,res) {
    res=str
    if       (str=="A") res="ALA"
    else if  (str=="C") res="CYS"
    else if  (str=="D") res="ASP"
    else if  (str=="E") res="GLU"
    else if  (str=="F") res="PHE"
    else if  (str=="G") res="GLY"
    else if  (str=="H") res="HIS"
    else if  (str=="I") res="ILE"
    else if  (str=="K") res="LYS"
    else if  (str=="L") res="LEU"
    else if  (str=="M") res="MET"
    else if  (str=="N") res="ASN"
    else if  (str=="P") res="PRO"
    else if  (str=="Q") res="GLN"
    else if  (str=="R") res="ARG"
    else if  (str=="S") res="SER"
    else if  (str=="T") res="THR"
    else if  (str=="V") res="VAL"
    else if  (str=="W") res="TRP"
    else if  (str=="Y") res="TYR"

    return res
}

function to_1_letter(res,str) {
    str=res
    if       (res=="ALA") str="A"
    else if  (res~/^CYS/) str="C"
    else if  (res=="ASP") str="D"
    else if  (res=="GLU") str="E"
    else if  (res=="PHE") str="F"
    else if  (res=="GLY") str="G"
    else if  (res=="HIS") str="H"
    else if  (res=="ILE") str="I"
    else if  (res=="LYS") str="K"
    else if  (res=="LEU") str="L"
    else if  (res=="MET") str="M"
    else if  (res=="ASN") str="N"
    else if  (res=="PRO") str="P"
    else if  (res=="GLN") str="Q"
    else if  (res=="ARG") str="R"
    else if  (res=="SER") str="S"
    else if  (res=="THR") str="T"
    else if  (res=="VAL") str="V"
    else if  (res=="TRP") str="W"
    else if  (res=="TYR") str="Y"
    else if  (res=="LYS+") str="K"
    else if  (res=="ARG+") str="R"
    else if  (res=="HIS+") str="H"
    else if  (res=="ASP-") str="D"
    else if  (res=="GLU-") str="E"

    return str
}