Bug fixes and final updates before RDRP docking.

config.py

@@ -3,7 +3,7 @@
 # database know what effective virtual host name you have.
 
 from pathlib import Path, PurePosixPath
-import os, subprocess
+import os, subprocess, time
 
 bucket = 'gs://ccddc'
 

create_inp.py

@@ -17,12 +17,30 @@ def fix(s):
         return m[0]
     return s[s.index("ROOT"):]
 
+# Pandas has too many types
+def get_name(df, i, idx):
+    if idx:
+        return df.index[i]
+    else:
+        return df.iloc[i]['name']
+
+# This script breaks out a dataframe into several
+# ligand pdbqt files.  It also prints a list of <name>\n<file>
+# pairs suitable for sending to autodock-GPU's filelist.
+#
+# Note the workarounds for different conventions of input
+# dataframes: frames with name-as-index
+# or pdbqt-s containing extraneous data to remove
 def main(argv):
     assert len(argv) == 2, "Usage: %s <ligs.pq>"
 
     df = pd.read_parquet(argv[1])
+    idx = True
+    if 'name' in df.columns:
+        idx = False
+
     for i in range(len(df)):
-        name = df.iloc[i]['name']
+        name = get_name(df, i, idx)
         conf = df.iloc[i]['conf']
         fname = name + '.pdbqt'
         with open(fname, "w") as f:

docker.sh

@@ -5,12 +5,12 @@
 #SBATCH --gres gpu:1
 #SBATCH -J dock
 #SBATCH -o %x.%A_%a.%j.out
-#SBATCH --array=1-866
+#SBATCH --array=1-868
 
 echo "Starting $SLURM_JOB_NAME-$SLURM_ARRAY_TASK_ID at" `date`
 source /apps/launchad/env.sh $SLURM_JOB_NAME
 
-export OMP_NUM_THREADS=2
+export OMP_NUM_THREADS=1
 srun -n2 -N2 --gres=gpu:1 --cpus-per-task=2 --exclusive \
      $LAD ccddc-controller $SLURM_JOB_NAME
 

helpers.py

@@ -76,7 +76,7 @@ def collect_lig(name, max_clusters=3):
     if len(xml) > max_clusters:
         xml = xml[:max_clusters]
 
-    return [(en, confs[i][1]) for en,i in xml], tors
+    return [(en, confs[i-1][1]) for en,i in xml], tors
 
 # inputstrings are all containing "DOCKED:"
 # [(en,pdbqt)]
@@ -134,7 +134,7 @@ def add_score(name, confs):
     conf, ntor = collect_lig(name, len(confs))
 
     for i in range(len(confs)):
-        if len(conf) < i:
+        if len(conf) <= i:
             confs[i][0].append(None)
             confs[i][1].append(None)
         else:

loadem.py

@@ -68,7 +68,8 @@ def run_job(job, item):
         ret = copyin(inp.format(**job))
         if ret: return ret
 
-    ret = subprocess.call( job['script'].format(**job), shell=True )
+    ret = subprocess.call( "set -e\n"
+			+ job['script'].format(**job), shell=True )
     if ret: return ret
 
     for out in job['out']:

small_docker.sh

@@ -5,7 +5,7 @@
 #SBATCH --gres gpu:1
 #SBATCH -J dock
 #SBATCH -o %x.%A_%a.%j.out
-#SBATCH --array=1-2
+#SBATCH --array=1-50
 
 echo "Starting $SLURM_JOB_NAME-$SLURM_ARRAY_TASK_ID at" `date`
 source /apps/launchad/env.sh $SLURM_JOB_NAME