From: Stefan Lyne Oliver (
Date: Wed Nov 09 2022 - 18:05:03 CST

Hi JC and Daipayan,

Apologies for the delayed response. I followed JC's suggestion and ran the tutorial on a couple of Linux iterations (I'll not go into the headaches this entailed but let's say it took far longer than it should have done!) and the tutorial works just fine. I ran this on both Centos8 and Ubuntu 22.04 WSL2 using VMD 1.9.4a55 and NAMD 2.14 multicore or NAMD 2.14 Linux-x86 64-multicore-CUDA. Here's a link to the output files from one Centos8...;!!DZ3fjg!_YnN1h2OUThqwgNCkChgZrW4nBU1lama2_cKOpJXvdjyZttIpUiPIs8QdBjwIWxfrZe9AbjtCy713TlMguSHxFo$ The issue does indeed seem to be with how VMD and NAMD are setup on Windows.

A couple of points. Firstly. it's not possible to perform the tutorial using VMD 1.9.3 because voltool is not an option. VMD 1.9.4 must be used. Secondly, I did have to spend a considerable amount of time troubleshooting NAMD. As stated previously, it won't run unless the mdff_template.namd file is in the same working directory (where the files are generated using VMD) as all the other files. From your troubleshooting, I understood this already and had copied the file into my working directory. However, both version of NAMD also failed if the par_all36_prot.prm was missing from the working directory. I'm not sure if this is an issue with how I used NAMD but it was consistent across both Centos8 and Ubuntu; NAMD did not function correctly by simply running the binary distribution as stated in the notes.txt file. In my experience, NAMD needs to have the step1.namd, step2.namd, mdff_template.namd, and par_all36_prot.prm all in the same working directory to successfully complete. Hopefully, that will help other users if the come across a similar scenario.

Thanks again for the troubleshooting. I got there eventually!



From: Daipayan Sarkar <>
Sent: Tuesday, 8 November, 2022 13:45
To: Gumbart, JC <>; Stefan Lyne Oliver <>
Subject: Re: vmd-l: RE: Simple MDFF Tutorial - 1ake does not fit into map

Huh. Admitedly, I never performed MDFF (or rather any MD) on a Windows OS workstation.
But based on my earlier response, as the MISC column in all latest logs are zero, no density-based grid forces are applied to selected atoms.

From: "Gumbart, JC" <<>>
Date: Tuesday, November 8, 2022 at 1:01 PM
To: Stefan Lyne Oliver <<>>
Cc: Daipayan Sarkar <<>>, "<>" <<>>
Subject: Re: vmd-l: RE: Simple MDFF Tutorial - 1ake does not fit into map

Interesting - I wonder if there's a bug in the Windows compilations of NAMD where gridForces don't actually get applied? Maybe someone in Urbana could test it?

In the meantime Stefan, I don't suppose you might have a Mac or even a Linux machine to use?


On Nov 8, 2022, at 12:49 PM, Stefan Lyne Oliver <<>> wrote:

Hi JC and Daipayan,

Thanks for taking the time to try and resolve my issues with MDFF. Firstly, apologies for not including the mdff_template.namd file; it was not included in the uploaded files. As per JC's suggestion a retried the simulation with the files I uploaded with a couple of different versions of NAMD (2.13 Win64-multicore-CUDA and 2.14_Win64-multicore). Neither NAMD version produced log files that contain values in the MISC column (attached), and, as you stated, do not fit into the density. Perhaps it's related to the mdff_template.namd file. My first question is, where does this file come from; is it included with the VMD installation? I noticed that the date of the mdff_template.namd file in my VMD installation folder is 20-Jun-21 (attached), which leads me to my second question is, should this file be modified in any way when the NAMD files are created? I'm puzzled as to where I've made the mistake. Thanks again for your continued help in troubleshooting this issue.



From: Daipayan Sarkar <<>>
Sent: Monday, 7 November, 2022 23:19
To: Gumbart, JC <<>>; Stefan Lyne Oliver <<>>
Subject: Re: vmd-l: RE: Simple MDFF Tutorial - 1ake does not fit into map

Hi Stefan,

Based on the link you shared and which JC nicely distributed in the email, I loaded the trajectory for "NAMD-step1" and see indeed the structure is not fitting into the density, which it totally should as JC pointed out. Next, I checked if the rigid body docking step was performed correctly, meaning if structure initially was "somewhat" in the density. In this case, that was true as well and simple MDFF should have worked. Then, next I checked your gridpdb file (atoms in the structure that will be fitted to the density) and that looks ok as well, as I believe you prepared this file with VMD atom selection "protein and noh".

Finally, I looked for the configuration script mdff_template.namd, which for some reason is not present in the same directory as you other NAMD files. This is strange as how you generated a trajectory, as clearly when you try to run NAMD, it should immediately bail out, with error saying that it cannot find mdff_template.namd. However, let's assume that this is a file upload issue. So, I went and looked in your NAMD log file "NAMD-step1/adk-step1.log", and this is the reason why the trajectory is not fitting into the density. Note, the MISC column in your log file, the column is all zero for the last step. If NAMD was applying forces to the selected atoms that are guided by the underlying EM density potential, then there should be a non-zero number in this column. This is not the case here, which by itself is very strange, unless the line was commented when performing MDFF. If the latter is true, then of course you are not performing any flexible fitting and running simple MD. Can you re-run the same input for step-1, with mdff_template.namd in the same directory and then track the MISC column in the generated to ensure that you get a non-zero value? I'll be happy to debug this with you further.

>>>> From NAMD-step1/adk-step1.log >>>>>>>

TIMING: 50000 CPU: 1187.49, 0.02369/step Wall: 1187.49, 0.023691/step, 0.00131617 hours remaining, 295.414063 MB of memory in use.

ENERGY: 50000 1080.1940 1747.9157 2045.9022 137.8512 -37.0483 -878.7146 0.0000 0.0000 2893.9367 6990.0368 290.5907 4096.1001 7014.7177 296.8822

PRESSURE: 50100 0 0 0 0 0 0 0 0 0
GPRESSURE: 50100 0 0 0 0 0 0 0 0 0
PRESSAVG: 50100 0 0 0 0 0 0 0 0 0
GPRESSAVG: 50100 0 0 0 0 0 0 0 0 0
ENERGY: 50100 1104.9235 1702.1816 2042.5091 149.5531 -35.9969 -938.5077 0.0000 0.0000 2974.7706 6999.4334 298.7075 4024.6628 7020.9036 298.3081

PRESSURE: 50200 0 0 0 0 0 0 0 0 0
GPRESSURE: 50200 0 0 0 0 0 0 0 0 0
PRESSAVG: 50200 0 0 0 0 0 0 0 0 0
GPRESSAVG: 50200 0 0 0 0 0 0 0 0 0
ENERGY: 50200 1153.4509 1709.9222 2061.5899 154.3728 -35.7245 -881.4146 0.0000 0.0000 2948.3775 7110.5742 296.0573 4162.1967 7134.7622 299.8366


WallClock: 1192.473022 CPUTime: 1192.473022 Memory: 295.437500 MB

>>>>> End of NAMD-step1/adk-step1.log >>>>>


From: "<>" <<>> on behalf of "Gumbart, JC" <<>>
Date: Tuesday, November 8, 2022 at 12:28 AM
To: Stefan Lyne Oliver <<>>
Cc: "<>" <<>>
Subject: Re: vmd-l: RE: Simple MDFF Tutorial - 1ake does not fit into map

Please keep the list copied in replies.

I went through the tutorial, generated the files, ran the simulation, and clearly see what's expected. I then ran the simulation with your files and again, I see what I expect. I can only think of a couple possible things: 1) there's a difference in our mdff_template.namd files (shouldn't be but you didn't have this in the files you provided) or 2) something is wonky with the version of NAMD you used (NAMD 2.14 for Win64-multicore-CUDA). Can you try with a different version of NAMD and see? It would be good to know if one of them has a bug.


On Nov 7, 2022, at 12:08 AM, Stefan Lyne Oliver <<>> wrote:

Thanks for the reply, JC. Here's a link to a shared folder that contains all of the files that were generated following the tutorial instructions...;!!DZ3fjg!_YnN1h2OUThqwgNCkChgZrW4nBU1lama2_cKOpJXvdjyZttIpUiPIs8QdBjwIWxfrZe9AbjtCy713TlMO6jGokY$ <;!!DZ3fjg!4-347YaJgeEY8KxGavMl-vhFsdJ89GOs0TEMJpv3ESheyUBtcCYInpQv3n2XauBU1Zuxyjwn84r68RyFcYHE3ynPEyI$>

There are also two subfolders for the NAMD-step1 and -step2 outputs.

Many thanks!


From: Gumbart, JC <<>>
Sent: Sunday, November 6, 2022 13:29
To: Stefan Lyne Oliver <<>>
Subject: Re: vmd-l: RE: Simple MDFF Tutorial - 1ake does not fit into map

Can you provide your files somewhere so others can take a look?


On Nov 4, 2022, at 7:52 PM, Stefan Lyne Oliver <<>> wrote:

Hi All,

I've still not come up with a solution for the problem below. Is there anyone who can advise?



From: Stefan Lyne Oliver
Sent: Tuesday, 1 November, 2022 20:10
Subject: Simple MDFF Tutorial - 1ake does not fit into map

Hi All,

I've been working through the simple MDFF tutorial. Unfortunately, after completing the NAMD (2.14 for Win64-multicore-CUDA) simulations the 1ake does not fit into the map. There is evidence that the molecular dynamics is working as I can see movement of the 1ake model after I've performed the NAMD simulation. I've tried this on two different windows machines (VMD 1.9.4a53) with reproducible results. The provided ake-step1-result.dcd and ake-step2-result.dcd files work fine when I load the molecules so I must be missing something in the tutorial. I've spent a fair amount of time on this and also discovered what seem to be some errors in the commands e.g. it looks like in the 'Defining secondary structure restraints' section the 1ake-initial_autopsf.psf file in step 2 should be 1ake-initial-docked_autopsf.psf. Correcting either way produces the same results. Can you please let me know what I'm missing in the tutorial so that I can be confident in using the procedure on my own maps and models. Thanks in advance.

Best wishes,


Stefan Oliver, Ph.D.<;!!DZ3fjg!-RcSuTdw9bWJ9ej-TyzBAd_siy68pglqwMic7u6tDtImmwMF1gfZK3oaJ24pWJk_r1btyGFzHIzORB9Rf7a6a5U$>
Dept. of Pediatrics, Stanford University School of Medicine, BMI Rm 2452, 240 Pasteur Drive, Stanford, California 94305.
We're enhancing STEM education for middle school and high school students in the Bay Area! Find out how you can get involved at;!!DZ3fjg!_YnN1h2OUThqwgNCkChgZrW4nBU1lama2_cKOpJXvdjyZttIpUiPIs8QdBjwIWxfrZe9AbjtCy713TlMD8kLypw$ <;!!DZ3fjg!-RcSuTdw9bWJ9ej-TyzBAd_siy68pglqwMic7u6tDtImmwMF1gfZK3oaJ24pWJk_r1btyGFzHIzORB9R2psSZPs$>

"We are so far from knowing all the agents of nature and their diverse modes of action that it would not be philosophical to deny phenomena solely because they are inexplicable in the actual state of our knowledge. But we ought to examine them with an attention all the more scrupulous as it appears more difficult to admit them." Laplace 1814