Hello all,
The charmm-gui absolute ligand binder (https://www.charmm-gui.org/?doc=input/afes.abinding) creates files for running FEP/λREMD simulations using NAMD. The default settings employs 32 λ -windows with general soft-core potentials and 10 consecutive FEP calculations (10 x 200 ps) are performed. The script I am using is shown below, edited for a bash environment but otherwise it is provided by charmm-gui.
# Absolute FEP
# Runing 2 ns (200 ps x 10)
export system=site
export cntmin=0
export cntmax=10
for cnt in $(eval echo "{$cntmin..$cntmax}")
do
echo $cnt
if [[ $cnt -eq 0 ]]; then
time -p mpirun -np 32 namd2 +ppn 4 +replicas 32 fep_${system}.conf --source FEP_remd_softcore.namd +stdout output_${system}/%d/job${cnt}.%d.log
else
time -p mpirun -np 32 namd2 +ppn 4 +replicas 32 restart_${cnt}.conf --source FEP_remd_softcore.namd +stdout output_${system}/%d/job${cnt}.%d.log
fi
done
Currently, each replica has 4 CPU cores assigned to them (128 cores total). I noticed after 4 hours, the calculation was still on the first FEP calculation out of 10 and the Benchmark time (updated only at the beginning at step 1000) showed 1.5 days. Assuming this is correct, 1.5 days (per FEP) x 10 would take 15 days. This seems very unreasonable. The most cores I can assign to each replica is 16 if I use 512 CPU cores which is costly.
Am I doing something wrong?
Has anyone used the charmm-gui absoulte ligand binder (https://www.charmm-gui.org/?doc=input/afes.abinding) with NAMD. How long should this type of calculation take? This is only for the complex. I would also run the ligand concurrently.
I haven’t used the ligand binder or carried out FEP/REMD in NAMD. But in my experience with NAMD I did not get the greatest benchmarks when running on CPUs only. The efficiency you report is consistent with my experience And I don’t notice anything obviously wrong with your code. If you have access to GPUs you could probably increase your speed. Amber is really fast but is optimized for GPUs.
- Badges
- Science topic
- Similar topics
- Biological Science
- Microbiology
- Virology
- Virus