#! /bin/bash
# Last edited on 2008-06-13 09:50:52 by stolfi

# Usage: get-region-lengths {SETNAME} {SETDIR} {BASDIR} {OUTDIR}
# Extracts region length statistics for a set of labeled genes. 

# Reads the gene names from the file "{SETDIR}/{SETNAME}.gset", and 
# the labeling of each gene from the file "{BASDIR}/{GENE}.lab"

# Tabulates the lengths of exon and non-exon regions
# in the given genes of Reese's dataset.  Writes the 
# statistical parameters (average length, etc.) and the 
# length histogram to the file "{OUTDIR}/{SETNAME}-{TYPE}.lens" 
# where {TYPE} is "K" or "N".

setName="$1"; shift
setDir="$1"; shift
basDir="$1"; shift
outDir="$1"; shift

setFile="${setDir}/${setName}.gset"
outName="${outDir}/${setName}"

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

cat ${setFile} \
  | gawk \
      -v maxLength=2000 \
      -v basDir="${basDir}" \
      '/^[a-zA-Z0-9]/{ printf "%s/%s.lab\n", basDir, $1; }' \
  > ${dataBank}

${progdir}/dbd_gather_region_length_stats -v outName=${outName} `cat ${dataBank}` 

/bin/rm -f ${dataBank}