Running specific monomer models on alphafold 2.2

[derf0]

Due to the long time required to generate structures of 3000+ residue proteins, I would like to be able to specify running only a specific model, such as model_1 or model_2 In earlier versions of alphafold this was possible with the option "--model_names=model_1,model_2" I am using alphafold on a machine where I am not root. I have also tried it on my installation on a machine with a small GPU, where I am running alphafold using docker, and making the following change in the model/config.py file from: MODEL_PRESETS = { 'monomer': ( 'model_1', 'model_2', 'model_3', 'model_4', 'model_5', ),

to: MODEL_PRESETS = { 'monomer': ( 'model_1', ),

still results in all five models being run. Thanks Fred

[epenning]

I did something similar, although I just added a new preset with my limited models. Your change looks like it should work though. These are the changes I made (this is for the monomer PTM single-model preset): https://github.com/epenning/alphafold/commits/main

I think you'll also have to make sure to do the Docker build again to make your change get picked up for running in Docker:

docker build -f docker/Dockerfile -t alphafold .

[derf0]

I have tried this out, and yes it works great on my server, where I am root and I am running alphafold, and I installed alphafold.

However, in order to generate structures of larger proteins than are possible on my little box, I am running alphafold on someone else's machine where I have to ask someone else to make the modifications, and they are hesitant.

I suggest that allowing selection of which model(s) to run as a command line option would be a nice addition to future versions of alphafold, and would make life much easier for the many people running alphafold who did not install it themself.