#!/bin/bash
#PBS -l walltime=01:00:00
#PBS -l ncpus=1,mem=2gb
#PBS -q uv2000
#PBS -N downstream_setup
#PBS -j oe


echo "PROJECT_ID = $PROJECT_ID"
echo "DATE = $DATE"
echo "BATCH = $BATCH"
DATE_BATCH=${DATE}_${BATCH}
echo "DATE_BATCH = ${DATE_BATCH}"


BASE=/scratch/u035/u035/shared/analysis/wes_pilot
SOURCE_DIR=/scratch/u035/u035/shared/analysis/wes_pilot/bcbio/final
PED_DIR=/scratch/u035/u035/shared/analysis/wes_pilot/params 


WORK_DIR=$BASE/${PROJECT_ID}
G2P_DIR=${WORK_DIR}/G2P
VASE_DIR=${WORK_DIR}/VASE 
COV_DIR=${WORK_DIR}/COV
DEC_DIR=${WORK_DIR}/DECIPHER
CNV_DIR=${WORK_DIR}/CNV
SCRIPTS_DIR=/home/u035/u035/shared/scripts

PYTHON2=/home/u035/u035/shared/software/bcbio/anaconda/envs/python2/bin/python2.7


# create the working dir and the required subfolders
mkdir ${WORK_DIR}
cd ${WORK_DIR}
mkdir PED
mkdir VCF
mkdir LOG
echo "Created WORK_DIR = ${WORK_DIR} for this batch and the required subfolders"


# create the folders needed for G2P
mkdir ${G2P_DIR}
cd ${G2P_DIR}
echo "Created ${G2P_DIR} for this batch"


# create the folders needed for VASE
cd ${WORK_DIR}
mkdir ${VASE_DIR}
cd ${VASE_DIR}
echo "Created ${VASE_DIR} for this batch"


# create the folders needed for the coverage analysis
cd ${WORK_DIR}
mkdir ${COV_DIR}
echo "Created ${COV_DIR} for this batch"


# create the DECIPHER folders
cd ${WORK_DIR}
mkdir ${DEC_DIR}
cd ${DEC_DIR}
mkdir IGV
echo "Created ${DEC_DIR} for this batch and the required subfolders"


# create the CNV folders
cd ${WORK_DIR}
mkdir ${CNV_DIR}
echo "Created ${CNV_DIR} for this batch"







### Copy the PED file per each family  ###
### format: <BATCH_ID>_<FAMILY_ID>.ped ###
###     and create the FAM_IDs file    ###

for FILE in ${PED_DIR}/${BATCH}_*.ped
do
    # copy the PED file
    cp $FILE ${WORK_DIR}/PED

#    # create the file with the family ids
#    filename="${FILE##*/}"
#    IFS='_|.' read -ra array_1 <<< "$filename"
#    echo "family id = ${array_1[1]}"
#    echo ${array_1[1]} >> ${WORK_DIR}/FAM_IDs.txt
done


######################################################################################
### generate the FAM_IDs.txt, PRO_IDs.txt and FAM_PRO.txt *only for trio* families ###
######################################################################################

time ${PYTHON2} ${SCRIPTS_DIR}/extract_trio_FAM_PRO_ID.py ${WORK_DIR}




echo "OK: Setup for PROJECT_ID = $PROJECT_ID (DATE_BATCH = ${DATE_BATCH}) successful"












######################################################

# LOG_DIR=$BASE/${PROJECT_ID}/logs
#mkdir ${LOG_DIR}
#echo "Created LOG_DIR = ${LOG_DIR} for this batch"

#cd $LOG_DIR
#echo "Created LOG_DIR = ${LOG_DIR} for this batch"


#cd $BASE
#mkdir $BASE/$PROJECT_ID




#source $CONFIG

## Change to the working directory
#cd $WORK_DIR/gvcfs

## Copy the GVCF files
#for file in `cat ../params/$BATCH.gvcfs.txt`
#do
#  cp $file* ./
#done

## Check the md5s
#for file in *.md5
#do
#  md5sum --check $file
#done