bench

A program to find frequent molecular substructures and discriminative fragments in a database of mol

#!/bin/bash

CLASSPATH=".."; export CLASSPATH

ulimit -t 1200 -v 1048576

mode=$1
if [[ $mode != "ext"     && $mode != "prune" \
   && $mode != "perfect" && $mode != "ring" ]]; then
  echo "usage: $0 (ext|prune|perfect|ring) [dataset [supplist [options]]]"
  exit
fi

test=${2:-"ic93"}
list=$3
if [[ $list == "" ]]; then
  if   [[ $test == "ic93" ]];
  then list="3.0 3.5 4.0 4.5 5.0 5.5 6.0";
  elif [[ $test == "steroids" ]];
  then list="-1 -2 -3 -4 -5 -6 -7 -8";
  else list="10 20 30 40 50 60 70 80 90 100";
  fi
fi
shift; shift; shift

data=data/$test.sln
base="java -Xmx800M moss.Miner -isln -osln ~ $data"

rm -f moss.sub $test.res

for supp in $list; do
  rm -f $test.tmp

  echo "support: $supp" | tee -a $test.tmp

  # --- comparison of extension types ---
  if [[ $mode == "ext" ]]; then
    cmd="$base -s$supp -e -p $*"
    echo "extension: rightmost" | tee -a $test.tmp
    $cmd -q -g 2>&1             | tee -a $test.tmp
    echo "extension: rgtm+can"  | tee -a $test.tmp
    $cmd +q -g 2>&1             | tee -a $test.tmp
    echo "extension: maxsrc"    | tee -a $test.tmp
    $cmd -q    2>&1             | tee -a $test.tmp
    echo "extension: msrc+can"  | tee -a $test.tmp
    $cmd +q    2>&1             | tee -a $test.tmp
  fi

  # --- comparison of pruning techniques 1 ---
  if [[ $mode == "prune" ]]; then
    cmd="$base -s$supp $*"
    echo "pruning: none"        | tee -a $test.tmp
    $cmd -e -p -q 2>&1          | tee -a $test.tmp
    echo "pruning: perfect"     | tee -a $test.tmp
    $cmd -e +p -q 2>&1          | tee -a $test.tmp
    echo "pruning: equiv"       | tee -a $test.tmp
    $cmd +e -p -q 2>&1          | tee -a $test.tmp
    echo "pruning: canonic"     | tee -a $test.tmp
    $cmd -e -p +q 2>&1          | tee -a $test.tmp
    echo "pruning: equ+can"     | tee -a $test.tmp
    $cmd +e -p +q 2>&1          | tee -a $test.tmp
    echo "pruning: per+can"     | tee -a $test.tmp
    $cmd -e +p +q 2>&1          | tee -a $test.tmp
    echo "pruning: all"         | tee -a $test.tmp
    $cmd +e +p +q 2>&1          | tee -a $test.tmp
  fi

  # --- comparison of pruning techniques 2 ---
  if [[ $mode == "perfect" ]]; then
    cmd="$base -s$supp -e +q $*"
    echo "pruning: none"        | tee -a $test.tmp
    $cmd -p -P 2>&1             | tee -a $test.tmp
    echo "pruning: partial"     | tee -a $test.tmp
    $cmd -p +P 2>&1             | tee -a $test.tmp
    echo "pruning: full"        | tee -a $test.tmp
    $cmd +p -P 2>&1             | tee -a $test.tmp
  fi

  # --- comparison of ring mining ---
  if [[ $mode == "ring" ]]; then
    cmd="$base -s$supp -r5:6 -e -p $*"
    echo "pruning: repos"       | tee -a $test.tmp
    $cmd -R -q 2>&1             | tee -a $test.tmp
    echo "pruning: close"       | tee -a $test.tmp
    $cmd -O +q 2>&1             | tee -a $test.tmp
    echo "pruning: merge"       | tee -a $test.tmp
    $cmd -E +q 2>&1             | tee -a $test.tmp
    echo "pruning: reorder"     | tee -a $test.tmp
    $cmd -R +q 2>&1             | tee -a $test.tmp
  fi

  gawk '
  ($1 == "support:")   { support = ($2 >= 0) ? $2 : -$2; }
  ($1 == "pruning:")   { prlist[n++] = pruning = $2; }
  ($1 == "extension:") { prlist[n++] = pruning = $2; }
  ((NF > 2) && ($(NF-2) == "substructure(s)]")) {
    subs[pruning] = substr($(NF-3), index($1, "[") +1);
    time[pruning] = substr($NF, 2, length($NF)-4);
  }
  ($1 == "number") && ($5 == "nodes")       { nodes [pruning] = $NF; }
  ($1 == "number") && ($4 == "fragments")   { frags [pruning] = $NF; }
  ($1 == "number") && ($4 == "embeddings:") { embeds[pruning] = $NF; }
  END {
    printf("%6.2f        ", support);
    printf("time     subs    nodes    frags    embeds\n");
    for (i = 0; i < n; i++) {
      pruning = prlist[i];
      printf("%-10s ", pruning);
      printf("%7.2f ", time[pruning]);
      printf("%8d ",   subs[pruning]);
      printf("%8d ",   nodes[pruning]);
      printf("%8d ",   frags[pruning]);
      printf("%9d\n",  embeds[pruning]);
    }
  }' $test.tmp >> $test.res
done

echo ""
cat $test.res

rm -f moss.sub $test.tmp