extract_results_from_benchmark.sh

#!/bin/bash

if [ -z "$5" ]; then
    echo "usage: $0 <dir with .results files> <dir for output> <min seq separation> <max seq separation> <target list file>" >&2
    exit 1
fi

#results_dir=${PWD}/psicov150/results
#out_dir=${PWD}/bench_results_aggregated

results_dir=$1
out_dir=$2
topomin=$3
topomax=$4
target_list=$5
out_master_file=${out_dir}/aggregate_results_min${topomin}_max${topomax}.txt

mkdir -p $out_dir

# print a header to the master file.
## TODO would be nice to have length in residues. It appears the pdbs are all generated by Modeller; this allows us to make some helpful assumptions.
## 	for n in *.pdb; do printf "$n "; grep ' CA ' $n | wc -l; done	

echo "pdbid actual_obs precL precL2 precL5 precL10 precLeq100 Lmax_prec_gt_0.5" > $out_master_file

for n in $(cat $target_list)
do
    tmp_file=$out_dir/${n}.tmp
    results_file=${results_dir}/${n}.results

    if [ ! -s $results_file ]
    then
	echo "$0: WARNING: $results_file not found or is empty. Skipping." >&2
	continue
    fi

    tail -n 7 $results_file > $tmp_file
    
    printf "$n " >> $out_master_file
    #extract various bits and put them in separate files.
    awk 'BEGIN {ORS=" "} { if (NF==5) {print $5} else {print $4} }' $tmp_file >> $out_master_file
    echo '' >> $out_master_file
done

rm $out_dir/*.tmp