search

openvax / mhcflurry

Peptide-MHC I binding affinity prediction
http://openvax.github.io/mhcflurry/
Apache License 2.0
191 stars 57 forks source link

MHC-flurry 1.6 use #161

Closed amomin-pact closed 4 years ago

amomin-pact commented 4 years ago

Hi , I was wondering if there are specific instructions to install and use MHCflurry 1.6. I see that algorithm has been modified to include peptide presentation efficiency.

Thanks Amin

timodonnell commented 4 years ago

Hi Amin - the 1.6.0 version is on pypi so if you pip install mhcflurry you should get that version. When you run mhcflurry-predict or mhcflurry-predict-scan as in the readme you will get antigen processing predictions as well as MHC affinities using this new version. Let us know if you hit any issues.

Tim

amomin-pact commented 4 years ago

I was able to install mhcflurry 1.6. How can I switch between 1.4 and 1.6 without switching conda environment?

Thanks

timodonnell commented 4 years ago

We don't have any builtin way to manage multiple installations of mhcflurry in the same environment. I think that'd be best handled using multiple conda environments.

However, if you're trying to use multiple versions of the mhcflurry predictive models, another option is to download all the models you want and then use the MHCFLURRY_DEFAULT_CLASS1_MODELS environment variable to switch between them (or just specify them directly when you load the predictor or call mhcflurry-predict). Any models released since the 1.2.0 version should be usable with the latest codebase, i.e. you can install mhcflurry 1.6.0 and then use the models that came out with 1.4.0 if you want by setting MHCFLURRY_DEFAULT_CLASS1_MODELS. It's a little complicated though since you'll need to download the models using 1.4.0, remember where they were saved, and then set the environment variable to point toward that. So multiple conda envs might still be simplest.