#! /bin/bash
# Last edited on 2008-06-13 08:49:27 by stolfi

# Usage: tabulate-probs {SETNAME} {WINSIZE} {LABEQV} {SETDIR} {BASDIR} {EVTDIR} {PRBDIR}

# Reads from the file "{SETDIR}/{SETNAME}.gset" a list of full item
# names, of the form "{SPECIES}/{ITEM}" where {SPECIES} is the name of
# a species's dataser in the EID database (e.g. "at2004", "hs35p1")
# and {ITEM} is an EID item identifier (e.g. "10008_NC_003074").

# Then reads the nucleotide files "{BASDIR}/{SPECIES}/{ITEM}.bas" and
# their labels "{BASDIR}/{SPECIES}/{ITEM}.lab" for all those items,
# and maps each label letter in the alphabet 'DEFHIJNXYPQRS' to the
# corresponding letter in the string {LABEQV}. Also maps all each
# lowecase nucleotide base to upper case, and replaces the 'U' base by
# 'T'.

# Then for every pair {e,t} where {e} is a {k}-event, {t} is a
# {k}-tuple, and {k = WINSIZE}, it counts the number of occurrences
# {#(e,t)} in all those items together, and estimates the conditional
# probability {Pr(t|e)} from those counts.

# These data are written ou to the file
# "{PRBDIR}/{SETNAME}-{WINSIZE}-{LABEQV}.prb", one pair per line, in
# the format "{e} {t} {#(e,t)} {Pr(t|e)}".

# If the file "{EVTDIR}/{WINSIZE}-{LABEQV}.evts" is present, it should
# contain a list of interesting {k}-events. In that case, this script
# only tabulates those pairs with events {i} in that list.
 
setName="$1"; shift
windowSize="$1"; shift
labEqv="$1"; shift
setDir="$1"; shift
basDir="$1"; shift
evtDir="$1"; shift
prbDir="$1"; shift

progdir="${STOLFIHOME}/programs/c/DNA/dnabayes"

setFile="${setDir}/${setName}.gset"
dataBank="/tmp/$$"
eventList="${evtDir}/${windowSize}-${labEqv}.evts"
if [[ -r ${eventList} ]] ; then
  eventOptions=( "-eventList" "${eventList}" )
else
  eventOptions=( "-all" )
fi
prbFile="${prbDir}/${setName}-${windowSize}-${labEqv}.prb"

# Create the bank file:
cat ${setFile} \
  | gawk \
      -v basDir="${basDir}" \
      '/[a-zA-Z0-9]/{ printf "%s/%s.bas %s/%s.lab\n", basDir, $1, basDir, $1; }' \
  > ${dataBank}

${progdir}/dbd_tabulate \
    -dataBank ${dataBank} \
    -windowSize ${windowSize} \
    -labelMap "DEFHIJNXYPQRS" = "${labEqv}" \
    ${eventOptions[@]} \
    -quiet \
    > ${prbFile}