# PMX repeatability is very poor
**URL:** https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234
**Category:** pmx
**Created:** [March 30, 2023, 10:31am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234 "2023-03-30T10:31:22Z")
**Posts on this page:** 9
**Page:** 1
### Author:  [@lyh](https://ask.bioexcel.eu/u/lyh)
#### Post date: [March 30, 2023, 10:31am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/1 "2023-03-30T10:31:22Z")
Dear Bioexcel users,
I use PMX to calculate when Ser mutates into Phe, when calculating ΔG in water, the repeatability is very poor, I use the version of gromacs2020.7.
In addition, I would like to ask if the trajectory file generated after using pmx can be used to analyze RMSD, RMSF and the number of hydrogen bonds? Because when I use pymol to view the mutation site, the key display of the mutation site is wrong, what should I do about this situation?
---
### Author:  [@vgapsys](https://ask.bioexcel.eu/u/vgapsys)
#### Post date: [March 30, 2023, 11:41am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/2 "2023-03-30T11:41:36Z")
What exactly do you mean with “poor repeatability”? What is the standard error that you get from multiple repeats? Is each individual simulation sufficiently converged?
Hybrid structures contain residues that can represent WT and Mut states. For A-state, you can just hide the dummies. For B-state, you would need to map the atoms to those of an actual structure in the B-state. If you omit the mutated residue, the rest of the structure follows the standard atom naming
---
### Author:  [@lyh](https://ask.bioexcel.eu/u/lyh)
#### Post date: [March 30, 2023, 12:56pm UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/3 "2023-03-30T12:56:06Z")
Thank you for your answer.
I repeated it a total of six times, with a standard error of 6.29, and each simulation was 40ns, checking that the RMSD already fluctuated very little.
---
### Author:  [@vgapsys](https://ask.bioexcel.eu/u/vgapsys)
#### Post date: [March 31, 2023, 8:40am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/4 "2023-03-31T08:40:37Z")
From these 5 runs I see that the standard error of the mean is ~2.5 kJ/mol which seems reasonable. Only water3 appears to be somewhat of an outlier. You could check visually if you spot anything different happening in that trajectory. Also, how long are your transitions?
---
### Author:  [@lyh](https://ask.bioexcel.eu/u/lyh)
#### Post date: [March 31, 2023, 10:40am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/5 "2023-03-31T10:40:47Z")
Thanks, I will use VMD to view the track file of this water3, in addition, I set the extraction 100 frames when unbalanced conversion, and the transitions time per frame is 100ps.
Finally, I would like to ask if PMX is compatible with gromacs 2020.7? I get an error when using the gmx rms command when parsing steps, but I get it right when using gromacs 5.1.4.
---
### Author:  [@vgapsys](https://ask.bioexcel.eu/u/vgapsys)
#### Post date: [March 31, 2023, 11:32am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/6 "2023-03-31T11:32:44Z")
pmx is compatible with gmx2020
Set your GMXLIB to the path with the mutation force fields to resolve these errors
---
### Author:  [@lyh](https://ask.bioexcel.eu/u/lyh)
#### Post date: [April 2, 2023, 2:05am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/7 "2023-04-02T02:05:06Z")
Yes, that’s how I set it up.
---
### Author:  [@vgapsys](https://ask.bioexcel.eu/u/vgapsys)
#### Post date: [April 3, 2023, 8:49am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/8 "2023-04-03T08:49:41Z")
Ok, would need to add dummy masses to the ff files.
Would it be an option to follow the suggestion that you see in the error description, i.e. use a .tpr file?
---
### Author:  [@lyh](https://ask.bioexcel.eu/u/lyh)
#### Post date: [April 4, 2023, 7:36am UTC](https://ask.bioexcel.eu/t/pmx-repeatability-is-very-poor/4234/9 "2023-04-04T07:36:15Z")
Yes, I tried to generate RMSD using a tpr file without error, but I found that using tp and using a gro file to generate RMSD was very different. My input file GRO is after the recovery track, while the TPR file is the original file that was generated, I will search for the cause of the problem.
