inital commit

Author: jhjensen2

highlight_atoms.py

@@ -0,0 +1,44 @@
+import os,sys
+from rdkit import Chem
+from rdkit.Chem import AllChem
+from rdkit.Chem import Draw
+from rdkit.Chem.Draw import IPythonConsole
+
+name2smiles = {}
+
+pka_file_name = sys.argv[1]
+smiles_file_name = pka_file_name.split("_")[0]+".smiles"
+svg_file_name = pka_file_name.split(".")[0]+".svg"
+
+smiles_file = open(smiles_file_name, "r")
+pka_file = open(pka_file_name, "r")
+
+for line in smiles_file:
+    words = line.split()
+    name = words[0]
+    smiles = words[1]
+    name2smiles[name] = smiles
+
+mols = []
+names = []
+atoms = []
+
+for line in pka_file:
+    words = line.split()
+    name = words[0]
+    smiles = name2smiles[name]
+    atoms.append(map(int,words[1][:-1].split(",")))
+    m = Chem.MolFromSmiles(smiles)
+    mols.append(m)
+    Chem.Kekulize(m)
+    names.append(name)
+    Draw.DrawingOptions.includeAtomNumbers=True
+
+img = Draw.MolsToGridImage(mols,molsPerRow=4,subImgSize=(200,200),legends=[x for x in names],useSVG=True,highlightAtomLists=atoms)
+
+
+print svg_file_name
+svg_file = open(svg_file_name, 'w')
+svg_file.write(img.data)
+svg_file.close()
+os.system('sed -i "s/xmlns:svg/xmlns/" '+svg_file_name)

merge_svg.py

@@ -0,0 +1,58 @@
+import sys
+
+def change_color(ellipse, name, find="#FF7F7F"):
+
+    red = "#e41a1c"
+    green = "#4daf4a"
+
+    color = green
+
+    if name == "red":
+        color = red
+
+    if find == "green":
+        find = green
+
+    ellipse = ellipse.replace(find, color)
+
+    return ellipse
+
+def merge_svg(svg_a, svg_b):
+
+#   svg_a = svg_a.split("\n")
+#   svg_b = svg_b.split("\n")
+
+    for i, a in enumerate(svg_a):
+        if "ellipse" in a:
+            svg_a[i] = change_color(a, "green")
+
+    for i, b in enumerate(svg_b):
+        if "ellipse" in b:
+            svg_b[i] = change_color(b, "green")
+
+    for i, b in enumerate(svg_b):
+        if b not in svg_a:
+            b = change_color(b, "red", find="green")
+            svg_a = svg_a[:i] + [b] + svg_a[i:]
+
+    svg_a = "\n".join(svg_a)
+
+    return svg_a
+
+
+if __name__ == "__main__":
+
+    svg_1_name = sys.argv[1]
+    svg_2_name = sys.argv[2]
+
+    svg_1 = open(svg_1_name, 'r').readlines()
+    svg_2 = open(svg_2_name, 'r').readlines()
+
+    svg = merge_svg(svg_1, svg_2)
+
+    output_name = svg_1_name.split(".")[0]
+    output_name = output_name[:-1]+"combined"
+    svg_file = open(output_name+".svg", 'w')
+    svg_file.write(svg)
+    svg_file.close()
+

submit_batches_mopac

@@ -0,0 +1,156 @@
+#!/bin/bash
+
+PARTITION=sauer # teach coms sauer
+NCPUS=8 # cores per job
+NBATCH=500 # input per batch
+EXTENSION=mop # extension to pack
+OUTEXT=out # extension to unpack
+
+# Check ssh key is setup
+ssh_pri=`ssh-keygen -y -f ~/.ssh/id_rsa`
+ssh_pub=`cat ~/.ssh/id_rsa.pub`
+if [[ $ssh_pub != $ssh_pri*  ]]; then
+
+    echo
+    echo "SSH key not setup. Please run 'ssh-keygen' in the terminal,"
+    echo "to setup a key-pair (ssh with no password), before running this"
+    echo "script"
+    echo
+
+    exit 4
+fi
+
+# note
+#   #SBATCH -w, --nodelist=asusnode103
+
+PWD=`pwd`
+JOB=${PWD##*/}
+SUBMIT=qsub.tmp
+host=$HOSTNAME
+
+
+files=`ls -f *$EXTENSION`
+
+# Remove input files which is already done
+out_files=`ls -f *.$OUTEXT 2> /dev/null`
+
+for x in $out_files; do
+    inp=${x%.*}.$EXTENSION
+    files=`echo $files | sed "s/$inp//"`
+done
+
+foo=( $files )
+nfiles=`echo "$files" | wc -w`
+
+echo $files
+echo $nfiles
+
+counter=0
+while [  $counter -lt $nfiles  ]; do
+
+    tarfile=batch_${counter}.tar.gz
+    outfile=batch_${counter}_log.tar.gz
+
+    tar -czf $tarfile ${foo[@]:$counter:$NBATCH}
+
+    cjob=$JOB\_\_$counter
+
+    cat > $SUBMIT <<!EOF
+#!/bin/bash
+#SBATCH --job-name=$cjob
+#SBATCH --nodes=1
+#SBATCH --cpus-per-task=$NCPUS
+#SBATCH --ntasks=1
+#SBATCH --error=$PWD/$cjob\_%j.stderr
+#SBATCH --output=$PWD/$cjob\_%j.stdout
+#SBATCH --time=100:00:00
+#SBATCH --partition=$PARTITION
+#SBATCH --no-requeue
+
+
+start=\`date +%s\`
+
+
+nodename=\$HOSTNAME
+echo 0 "Running on \$nodename"
+echo 0 "un-tar input files"
+
+
+mkdir /scratch/\$SLURM_JOB_ID
+cd /scratch/\$SLURM_JOB_ID
+
+
+cp $PWD/$tarfile .
+tar -xzf $tarfile
+
+# MOPAC
+
+cp -r /opt/mopac .
+cp -r /opt/intel/composer_xe_2013.1.117/compiler/lib/intel64 .
+export LD_LIBRARY_PATH=/scratch/\$SLURM_JOB_ID/intel64:$LD_LIBRARY_PATH
+export MKL_NUM_THREADS=1
+
+run_calculation () {
+
+    input=\${1%.*}
+    echo \$input
+    /scratch/\$SLURM_JOB_ID/mopac/MOPAC2016.exe \$input > /dev/null 2> /dev/null
+
+}
+
+export -f run_calculation
+
+# END MOPAC
+
+
+end=\`date +%s\`
+runtime=\$((end-start))
+echo \$runtime "run calculation"
+
+
+ls -f *.$EXTENSION | parallel -j$NCPUS "run_calculation {}"
+
+
+end=\`date +%s\`
+runtime=\$((end-start))
+echo \$runtime "tar output"
+
+
+tar -czf $outfile *.$OUTEXT
+cp $outfile $PWD
+
+
+end=\`date +%s\`
+runtime=\$((end-start))
+echo \$runtime "untar in $PWD"
+
+
+ssh sunray "cd $PWD && tar xzf $outfile"
+
+
+end=\`date +%s\`
+runtime=\$((end-start))
+echo \$runtime "done"
+
+
+convertsecs() {
+    ((h=\${1}/3600))
+    ((m=(\${1}%3600)/60))
+    ((s=\${1}%60))
+    printf "%02d:%02d:%02d\n" \$h \$m \$s
+}
+
+convertsecs \$runtime
+
+!EOF
+
+    # submit batch
+    sbatch $SUBMIT
+
+    # next
+    let counter+=$NBATCH
+
+done
+
+
+