Adaptative MD on local gpu

[vas2201]

Hello,
I am running the Jupyter notebook on a local GPU (not submitting jobs), and when I use the command "acemd3 input" inside the folder of one of the adaptive MDs, it works.
However, when I use the following command, I don't get any output:
local = LocalGPUQueue()
local.submit('./prod/')
local.wait()
Can you please suggest how I can run the adaptiveMD without submitting a job on a local GPU?.

Thanks
Vas

[stefdoerr]

is there a run.sh file in the ./prod/ folder? what does it contain?

[vas2201]

Please discard the above message to avoid confusion.
I attached the jpg of the jupyter notebook (adaptive protocol) as a drive link for your check. I am doing an adaptive MD on target RNA on local gpu (not submitting the job on hpc). Looks like it is running, but there is no log file ot xtc file in the directory (adaptivemd/input/e1s9_generators).

Could you please suggest to me to fix the problem?

Regards
Vas

google drive link : jupyter notebook

---

https://drive.google.com/file/d/1rtmgQ-LCwjLcrWPl3kK-rFRi2SyZemp_/view?usp=share_link

run.sh

---

#!/bin/bash
acemd3 >log.txt 2>&1

output :

---

ad.run()
2023-02-21 18:14:20,572 - htmd.adaptive.adaptive - INFO - Processing epoch 0
2023-02-21 18:14:20,572 - htmd.adaptive.adaptive - INFO - Epoch 0, generating first batch
2023-02-21 18:14:20,576 - htmd.adaptive.adaptive - INFO - Generators folder has no subdirectories, using folder itself
2023-02-21 18:14:21,701 - jobqueues.util - INFO - Trying to determine all GPU devices
2023-02-21 18:14:21,741 - jobqueues.util - INFO - GPU devices requested: 0
2023-02-21 18:14:21,741 - jobqueues.util - INFO - GPU devices visible: 1
2023-02-21 18:14:21,742 - jobqueues.localqueue - INFO - Using GPU devices
2023-02-21 18:14:21,743 - jobqueues.util - INFO - Trying to determine all GPU devices
2023-02-21 18:14:21,765 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s1_generators
2023-02-21 18:14:21,768 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s2_generators
2023-02-21 18:14:21,770 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s3_generators
2023-02-21 18:14:21,772 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s4_generators
2023-02-21 18:14:21,774 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s5_generators
2023-02-21 18:14:21,776 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s6_generators
2023-02-21 18:14:21,778 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s7_generators
2023-02-21 18:14:21,780 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s8_generators
2023-02-21 18:14:21,782 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s9_generators
2023-02-21 18:14:21,784 - jobqueues.localqueue - INFO - Queueing /home/nmrbox/spenumutchu/Desktop/Projects/aufr12_project/SL2_WT/hmtd_sl2wt/adaptive_md/adaptivemd/input/e1s10_generators
2023-02-21 18:14:21,786 - htmd.adaptive.adaptive - INFO - Sleeping for 120 seconds.
2023-02-21 18:16:21,887 - htmd.adaptive.adaptive - INFO - Processing epoch 1
2023-02-21 18:16:21,888 - htmd.adaptive.adaptive - INFO - Retrieving simulations.
2023-02-21 18:16:21,889 - htmd.adaptive.adaptive - INFO - 10 simulations in progress
2023-02-21 18:16:21,889 - htmd.adaptive.adaptive - INFO - Sleeping for 120 seconds.
2023-02-21 18:18:21,992 - htmd.adaptive.adaptive - INFO - Processing epoch 1
2023-02-21 18:18:21,992 - htmd.adaptive.adaptive - INFO - Retrieving simulations.
2023-02-21 18:18:21,993 - htmd.adaptive.adaptive - INFO - 10 simulations in progress
2023-02-21 18:18:21,993 - htmd.adaptive.adaptive - INFO - Sleeping for 120 seconds.
2023-02-21 18:20:22,091 - htmd.adaptive.adaptive - INFO - Processing epoch 1

[vas2201]

The issue is resolved, once updated the following script.
queue = LocalGPUQueue()
queue.datadir = './data'
queue = LocalGPUQueue()
queue.devices = [1,2,4]
ad = AdaptiveMD()
ad.app = queue
ad.nmin = 5
ad.nmax = 10
ad.nepochs = 30
I can see the log files and and xtc files the simulation in /input/e1s1_sl2_10ns_1/ folder (single gpu) now.

I use the following command (All Proton distances) to construct MSM.

protsel = 'nucleic and name H**'
ad.projection = MetricSelfDistance(protsel)
#ad.projection = MetricRmsd(trajrmsdstr='nucleic and noh', refmol=Molecule('generators/structure.pdb'))

Does name H** include all the proton distances ?

Thanks for your help.
Regards
Vas

[stefdoerr]

hmmm I'm not sure what H** does in atomselect. You want to get the distances of all hydrogens to each other in your nucleic acid? If yes I'd do nucleic and element H. But keep in mind this will create an insanely large distance matrix. If you have i.e. 500 hydrogens this is 500*499/2=124750 distances for each simulation frame. You will run out of memory very fast and crash your computer. You probably need to select some better metric for your conformations, like the distances of some backbone atom or of some specific protons (like one per residue only)

[vas2201]

Thank you.