Estimated Oscillation Time period less than timestep

User discussions
, ,
1

GROMACS version: 2022.1
GROMACS modification: No
Hi,
I was doing a Free energy calculation where I am mutating a residue from Alanine to Phe. During my production run, I am gettting the following error:

The bond in molecule-type Protein_chain_B between atoms 144 DCG and 145
DCD1 has an estimated oscillational period of 1.8e-02 ps, which is less
than 10 times the time step of 2.0e-03 ps.
Maybe you forgot to change the constraints mdp option.
I understand that I can remove this error by using a time step of <= 1.8 fs. But I would like to know why I am getting this error?
I checked the topology file and the bond parameters are same as that of two aromatic carbons.

Normal Phenylalanine has the following bond parameters:
CA CA 1 0.1375 255224.0

And my hybrid topology file has the following

144 DUM_CA 8 A2F DCG 140 0.000000 3.9636 CA 0.000000 12.0110
145 DUM_CA 8 A2F DCD1 141 0.000000 3.9636 CA -0.115000 12.0110

and the bonds section as this
144 145 1 0.137500 255224.000000 0.137500 255224.000000

Based on these values I calculated the oscillation time period manually and I believe it comes to be around 96.4 fs. If this is correct then why do gromacs calculate it as 18 fs? And if it 18 fs then why this particular bond? Why not bonds in a normal Phe or the other aromatic bonds in the hybrid residue? I also tried with gromacs2023.2 and the problem is still there. I’m skeptical that I may have missed something in the parameter files.

2

From my calculation it would seem that the mass of DUM_CA is 1. I think you would want this to be identical to the non-dummy mass of the same atom, which would also solve this issue.

3

Hi,

Can you please tell me which column you are referring to? I’m not sure I see where the mass of the DUM_CA is 1. I do see the mass of DUM_CA as 3.9636 but that is for all dummy Carbon atoms not specific to CA. For example, When I change from Leu to Val, the new DCG1/2 also has that same mass when it is a dummy atom and then in the lambda=1 section of the topology it is 12.011
I thought when it uses lambda=1, the DUM_CA has the mass of CA which is 12.011 which is what it morphs into.

4

Ah, yes, it’s 4. I now also see that in your first post.

It is in general better to avoid changing masses when you can. Simply use 12.0110. Why do you want to use 3.9636?

5

Hi,

I used pmx to automate the generation of hybrid topplogy and it put that value in dummy Carbon atoms. I can change those to 12.011 and see if that works. How does that mass matter, if you don’t mind me asking? Does this mean that if I am doing a slow growth method the DUM_CA starts with a mass of 4?
Also would it be possible for you to share some resources on how to generate hybrid topologies manually for example if I have an unnatural residue to which I would like my amino acid to mutate to.

Thanks