On FASPR and rotamer sampling

This forum is shown on the index page along with all topics.

Moderator: robpearc

zcx@umich.edu
Posts: 69
Joined: Wed Nov 03, 2021 5:59 pm

On FASPR and rotamer sampling

Post by zcx@umich.edu »

From: <l*@opnbnch.com>
Date: Friday, June 2, 2023 at 14:25
Subject: On FASPR and rotamer sampling
Hi Prof. Zhang,
I've been really impressed with using your FASPR tool - it's very fast and generates plausible coordinates. It's been a great addition to the suite of protein structure modeling tools out there at the moment - thank you for making it available!

I'm interested in generating ensembles of conformers with alternate rotamer states. One way to do this might be molecular dynamics / monte carlo, but I wondered if the FASPR tool can be hacked to do this. I looked in the code and found mention of a 'temp' variable, suggesting there's some kind of annealing going on. But the outputs are very consistent, suggesting that no sampling is occurring.

Do you know of a way to generate alternate conformers with FASPR? I'm happy to do a bit of hacking to implement it. Or alternatively, do you have a similar tool that might achieve the same end goal?

Thanks a lot,
zcx@umich.edu
Posts: 69
Joined: Wed Nov 03, 2021 5:59 pm

Re: On FASPR and rotamer sampling

Post by zcx@umich.edu »

FASPR aims to achieve the global minimum energy configuration (GMEC) by fully deterministic algorithms, such as dead end elimination. Therefore, it is not possible to force it to generate multiple conformation. The temp variable in Search.cpp is not related to temperature annealing.
xiaoqiah
Posts: 6
Joined: Tue May 04, 2021 5:05 pm

Re: On FASPR and rotamer sampling

Post by xiaoqiah »

Dear User,

As Chengxin (zcx@umich.edu) mentioned, a critical feature of FASPR is that it uses a hybrid, deterministic algorithm to search the global minimum energy configuration (GMEC) determined by the FASPR score function on the ratameric space defined by the Dunbrack 2010 rotamer library. Therefore, you can see that your repacked structures from different runs are always consistent.

I understand that you are interested in generating ensembles of conformers with alternate rotamer states, and this is also one of my interests. I developed some other Monte Carlo-based tools that may satisfy your needs, and if not, they can be easily hacked to do what you mentioned. Here are the two tools: (1) EvoEF2 (https://zhanggroup.org/EvoEF/, https://github.com/tommyhuangthu/EvoEF2). If you want to repack your protein, you can run a command like: ./EvoEF2 --command=ProteinDesign --pdb=your.pdb --wildtype_only; and (2) UniDesign (https://github.com/tommyhuangthu/uniDesign), which is extended from EvoEF2. You can run a similar command: ./EvoEF2 --command=ProteinDesign --pdb=your.pdb --wildtype_only --ntraj=N. The option "--ntraj=N" is used to generated N structures which may or may not be identical. If they are not identical, you may treat them like ensembles of conformers with alternate rotameric states.

Xiaoqiang Huang (xiaoqiah@umich.edu)
Post Reply