split_test.sh

#!/bin/bash

testfile=`realpath $1`
FEATURE=`echo $1 | xargs basename | sed -re 's/\.test\.tsv$//'`
CL=`echo $FEATURE | sed -re 's/^([^.]+)\..+$/\1/'`
if [ "$CL" == "all" ]; then
  DIRNAME=`echo $1 | xargs dirname | xargs --replace echo {}/.. | xargs realpath`
  train_file="${DIRNAME}/train/${FEATURE}.train.tsv"
  fit_file="${DIRNAME}/fit_chr_not2_9_22/${FEATURE}.fit_chr_not2_9_22.tsv"
  validation_file="${DIRNAME}/val_chr_2_9_22/${FEATURE}.val_chr_2_9_22.tsv"
  ladder_file="${DIRNAME}/ladder/${FEATURE}.ladder.tsv"
else
  DIRNAME=`echo $1 | xargs dirname | xargs --replace echo {}/../.. | xargs realpath`
  train_file="${DIRNAME}/train/${CL}/${FEATURE}.train.tsv"
  fit_file="${DIRNAME}/fit_chr_not2_9_22/${CL}/${FEATURE}.fit_chr_not2_9_22.tsv"
  validation_file="${DIRNAME}/val_chr_2_9_22/${CL}/${FEATURE}.val_chr_2_9_22.tsv"
  ladder_file="${DIRNAME}/ladder/${CL}/${FEATURE}.ladder.tsv"
fi

echo "paste /home/ivavlakul/dream/annotation/test_regions.blacklistfiltered.header.bed $testfile | grep -wEv '(chr1|chr8|chr21)' | cut -f4- > $train_file"
echo "paste /home/ivavlakul/dream/annotation/test_regions.blacklistfiltered.header.bed $testfile | grep -wEv '(chr1|chr8|chr21|chr2|chr9|chr22)' | cut -f4- > $fit_file"
echo "paste /home/ivavlakul/dream/annotation/test_regions.blacklistfiltered.header.bed $testfile | grep -wE '(chrom|chr2|chr9|chr22)' | cut -f4- > $validation_file"
echo "paste /home/ivavlakul/dream/annotation/test_regions.blacklistfiltered.header.bed $testfile | grep -wE '(chrom|chr1|chr8|chr21)' | cut -f4- > $ladder_file"