-
Notifications
You must be signed in to change notification settings - Fork 25
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Espaloma bonded terms discrepancies #209
Comments
Also adding the system XML files here. |
@laurenwinkler Hi, thank you for your interest in using Espaloma. I can reproduce your results using both RDkit and Openeye as cheminformatics backends. As far as I can tell, Espaloma is behaving as expected, since the protein sdf has 2632 bonds, and that's exactly what we get in the system XML file for the full system (complex with Espaloma). Considering the extra 21 bonds from the ligand file, of course. That makes me suspicious about the |
Sorry, never mind. I got confused. |
Thank you for you response. I have attached the protein pdb file as well as the script used to convert the pdb to an SDF. Would you mind sharing the script you used to produce the results as well as the XML file? |
@laurenwinkler I think the zip file wasn't successfully uploaded, can you retry uploading it? Thanks! |
My apologies! |
Just following up, could you share the script that you guys used to produce an XML file that had the correct bond terms for a complex with both protein and ligand described by Espaloma and no OE backend? @ijpulidos @kntkb If you look at the XML files in Lauren's first comment (system-xml-files) the complex system (espaloma-full-system.xml) has 2653 bonds where as amber-gaff.xml and espaloma-lig-only.xml have 1323 bonds. If you have a way to start from the PDB file and have this work, our issue may be solved. |
The extra bonds in I also realized that Although this is not elegant, a workaround could involve writing a parser to transfer the There is a sample script here that starts from a PDB file but requires OE in the backend. |
@kntkb Great catch! Yes, I agree, this explains the issue. Maybe what we want is for the |
Thank you for your help. We redid the XML file generation for espaloma and amber and now they have the same bond count. I've attached the system XML files as well as our scripts and inputs. Would you be able to verify if the XML files look correct to you? |
Not sure this is related, but in the hope it helps, it may be worth testing openmmforcefields |
@diogomart Ah, yes. The asymmetric charges are expected when using openmmforcefields |
Thanks for the pointers @diogomart @kntkb
We would love to get to the bottom of if Espaloma is able to run with rdkit under the hood so we can benchmark. If you get the chance to check out these XML files, we would like to know if you still see an issue with them? If you don't see any problems, this issue is resolved on our end. |
If the intention was to parameterize the system using |
The proteins are parameterized with As mentioned earlier, I believe you could use RDKit as an alternative, but you would need to use SDF format for the protein and transfer the information to after creating the XML file, as mentioned earlier. |
Hi Espaloma Devs,
I am exploring the use of Espaloma (with RdKit under the hood) to describe a protein + ligand complex.
When comparing system XML files from an (unsolvated) Espaloma described system vs one described by amber/protein.ff14SB.xml and GAFF, I noticed that ligands described with Espaloma contain more than (ligand count) choose 2 additional number of bonded terms in the section of the output XML file. For example, if the ligand has let's say 12 atoms, the XML of a system where the ligand is parameterized with Espaloma has:
additional_number__terms > 3 x (12 choose 2)
If the ligand described with Espaloma is in a complex, is it intentional to add nonphysical harmonic bonds between the ligand and the protein?
I've attached the script used to build the Espaloma systems as well as the input files. I should also note that I am using:
openff-interchange=0.3.18
openff-toolkit=0.14.3
openmm=8.0.0
espaloma=0.3.2
rdkit=2023.03.3
python=3.10
Any insight would be helpful. Thanks in advanced.
Espaloma-test.zip
The text was updated successfully, but these errors were encountered: