Closed spalgit closed 2 years ago
Hmm odd… the only thing I can imagine that would affect that is the ncpu parameter passed in the pyscreener
config. What does your config look like?
Hi David, Thanks. My ini and config file are. It is pretty much default. I am tying to give a ncpu=20 but it still runs on 8 cpus. Not sure why>
Thanks Sandeep
screen-type = vina
metadata-template = {"software": "vina"}
receptors = [proteinH.pdb]
center = [20.556, -14.266, 1.286]
size = [17, 10, 22]
ncpu = 20
------ And
config file is
[general]
output-dir = molpal_eMols_0.005_rf_ucb
--retrain-from-scratch yes
[pool]
library = libraries/eMols_similarity_unique.smi.gz
# no fps file will force MolPAL to write a new HDF5 file with the fingperprints
[encoder]
# the default encoder is Atom-pair of length 2048, min_path=1, max_path=3
[model]
# by default, we use an RF model
[acquisition]
# by default, we acquire inputs greedily
[objective]
# there are no default objective values
objective = docking
objective-config = examples/objective/docking_brpf1.ini
--minimize
[stopping]
# by default, MolPAl will explore until the fractional difference betewen the
# current top-k average and the moving average of the 3 recent top-k averages
# is less than 0.01. the default k value is equal to 0.05% of the pool size
Hmm yeah that looks correct. When you use vina by itself with 20 CPUs (by itself and inside pyscreener), do you observe full utilization?
No I do not observe full utilization. It uses only 8 cpus. I cannot understand why?
I like the results though.
just to confirm, you observe underutilization when you run the following command:
vina -r RECEPTOR -l LIGAND --ncpu=20 [...]
closing this due to inactivity
Hello,
I am using molpal with docking and it works well. I am using a subset of eMolecules library (~1M compounds) and trying to do a VS against a target. I am using a --init-size 0.005 and --batch-size 0.005.
The only thing I am not able to control is the number of CPUs on which the vina docking works. I am using ray start --head and my VM has 24 cpus and gpus. MolPAl only uses 8 out ot the 24 CPUs.
This problem is not there when I use pyscreener directly and it utilizes all the cpus. Can you please suggest which parameter should I change to make the docking explorations run on all 24 cpus?
Thanks Sandeep