mdtraj / nma Goto Github PK

This would allow a user to fit ANMs over an entire trajectory more quickly, only needing to update the Hessian when there is a new contact or one is lost.

Hi, we are attempting to compute the NMA of arrestin structures (e.g., PDB code 1JSY), each of which have around 3,000 atoms. We were able to make the code work on the provided water examples, but are having difficulty with larger molecules. In the resultant trajectories, a very small subset of the molecule seems to move significantly, while the rest of the molecule appears to be rigid. This makes us think that the code may only be looking at that small subset, but we have been unable to determine why this might be the case or if there is a different problem we’re not seeing. We would greatly appreciate some insight!

Suggested by @joegomes.

I'm thinking the best way to implement this is to treat the topology as a graph and find the closest bonded atom within the selection from which to inherit modes.

Just a heads up to any potential users, during installation Python might complain about a non-ASCII character in the setup.py file. This is resolved by replacing the accented a in "Hernandez" with a regular "a".

For larger proteins, searching for the nearest selected atom can take a while, so it's probably worth saving these in the ANMA object.