#!/usr/bin/tclsh
#
# This script was provided by Gabriel Cornilescu:
# http://www.nmrfam.wisc.edu/directory.html
# G.C. 2012

if {$argc != 2} \
   {
    puts stderr ""
    puts stderr "usage: "
    puts stderr "bmrb2talos.tclsh file.str type"
    puts stderr ""
    puts stderr "where 'type' is 2 or 3 for input in format NMR-STAR v2.1 or v3.1, respectively"
    puts stderr ""
    exit
   }

set inFile [lindex $argv 0]
set inFileID [open $inFile r]

set type [lindex $argv 1]

if {$type == 2} \
   {
# these work with BMRB's nmrstar2.1 format (e.g. bmr6011.str) and PINE:
    set resID_col 2
    set resType_col 3
    set atName_col 4
    set shift_col 6
    set seqFlag "_Mol_residue_sequence"
    set csFlag "_Chem_shift_ambiguity_code"
   } \
elseif {$type == 3} \
   {
# these work with BMRB's nmrstar3.1 format (e.g. bmr15000.str, bmr17769.str)
    set resID_col 5
    set resType_col 6
    set atName_col 7
    set shift_col 10
    set seqFlag "_Entity.Polymer_seq_one_letter_code"
    set csFlag "_Atom_chem_shift.Assigned_chem_shift_list_ID"
       } \
else \
   {
    puts stderr "\nEdit the script to input correct column numbering in the BMRB input file, if needed!\n"

# set manually the column numbers below (col. numbering starts with 0)
# for a different str (BMRB) format and uncomment the lines:
#    set resID_col x
#    set resType_col x
#    set atName_col x
#    set shift_col x
   }

#
# sequence offset number (to match pdb sequence, if needed):

set match 0

#
# atom type assignments needed for TALOS+:

set list "N H HN HA HA1 HA2 HA3 CA CB C"

while { [eof $inFileID] == 0} \
   {
    gets $inFileID line
    set test [lindex $line 0]

    if {$test == $seqFlag} \
       {
        gets $inFileID line
        set count 0
        set end 0
	set dlist ""

        while { $end == 0 } \
           {
            gets $inFileID line

            set first [string trim [lindex $line 0] ]
            set seq_1 [string trim [string range $line 0 9] ]
            set seq_2 [string trim [string range $line 10 end] ]

            if {$first != ";"} \
               {
                incr count

                lappend seqList $seq_1

                if {$seq_2 != ""} \
                   {
                    lappend seqList $seq_2
                   }

		append dlist "[string length [string trim $line]] "
               }

            if {$first == ";"} \
               {
                set end 1
               }
           }

        set last  [lindex $dlist [expr [llength $dlist] - 1]]

        set len   [expr ceil($last/10.0)]

        if {$len == 1} \
           {
            set string_no [expr $count*2 - 1]
           } \
        elseif {$len == 2} \
           {
            set string_no [expr $count*2]
           }

        set final [expr ceil($string_no/5.0)]

        for {set i 0} {$i < $final} {incr i} \
            {
             set j [expr $i*5]
             set row [expr $j + 4]

             puts "DATA SEQUENCE [lrange $seqList $j $row]"
            }
        }

    if {$test == $csFlag} \
       {
        puts ""
        puts "VARS RESID RESNAME ATOMNAME SHIFT"
        puts "FORMAT %4d %1s %4s %8.3f"
        puts ""

        gets $inFileID line
        gets $inFileID line

#puts "line=$line"

        while {[gets $inFileID line] > 0} \
           {
            set resID [expr [lindex $line $resID_col] + $match]
            set resType [lindex $line $resType_col]
            set atName [lindex $line $atName_col]

            if {$atName == "H"} {set atName HN}

            set shift [lindex $line $shift_col]

            switch $resType \
               {
                "ALA" {set resType "A"}
                "ARG" {set resType "R"}
                "ASN" {set resType "N"}
                "ASP" {set resType "D"}
                "CYS" {set resType "C"}
                "GLN" {set resType "Q"}
                "GLU" {set resType "E"}
                "GLY" {set resType "G"}
                "HIS" {set resType "H"}
                "ILE" {set resType "I"}
                "LEU" {set resType "L"}
                "LYS" {set resType "K"}
                "MET" {set resType "M"}
                "PHE" {set resType "F"}
                "PRO" {set resType "P"}
                "SER" {set resType "S"}
                "THR" {set resType "T"}
                "TRP" {set resType "W"}
                "TYR" {set resType "Y"}
                "VAL" {set resType "V"}
               }

            if {[lsearch $list $atName] >= 0 && $shift != "-"} \
               {
                puts [format "%5s %3s %-3s %9.3f" $resID $resType $atName $shift]
               }
           }
       }
   }

exit