Skip to main content

MHC Binding Predictor

Project description

Build Status Coverage Status


Open source neural network models for peptide-MHC binding affinity prediction

The adaptive immune system depends on the presentation of protein fragments by MHC molecules. Machine learning models of this interaction are used in studies of infectious diseases, autoimmune diseases, vaccine development, and cancer immunotherapy.

MHCflurry supports Class I peptide/MHC binding affinity prediction using ensembles of allele-specific models. You can fit MHCflurry models to your own data or download models that we fit to data from IEDB and Kim 2014. Our combined dataset is available for download here.

We are working on a performance comparison of these models with other predictors such as netMHCpan, which we plan to make available soon.

Pan-allelic prediction is supported in principle but is not yet performing accurately. Infrastructure for modeling other aspects of antigen processing is also implemented but experimental.


Install the package:

pip install mhcflurry

Then download our datasets and trained models:

mhcflurry-downloads fetch

From a checkout you can run the unit tests with:

nosetests .

The MHCflurry predictors are implemented in Python using keras.

MHCflurry works with both the tensorflow and theano keras backends. The tensorflow backend gives faster model-loading time but is undergoing more rapid development and sometimes hits issues. If you encounter tensorflow errors running MHCflurry, try setting this environment variable to switch to the theano backend:

export KERAS_BACKEND=theano

You may also needs to pip install theano.

Making predictions from the command-line

$ mhcflurry-predict --alleles HLA-A0201 HLA-A0301 --peptides SIINFEKL SIINFEKD SIINFEKQ

The predictions returned are affinities (KD) in nM. The prediction_low and prediction_high fields give the 5-95 percentile predictions across the models in the ensemble.

You can also specify the input and output as CSV files. Run mhcflurry-predict -h for details.

Making predictions from Python

>>> from mhcflurry import Class1AffinityPredictor
>>> predictor = Class1AffinityPredictor.load()
>>> predictor.predict_to_dataframe(peptides=['SIINFEKL'], allele='A0201')

  allele   peptide   prediction  prediction_low  prediction_high
  A0201  SIINFEKL  6029.084473     4474.103253      7771.297702

See the class1_allele_specific_models.ipynb notebook for an overview of the Python API, including fitting your own predictors.

Details on the downloadable models

An ensemble of eight single-allele models was trained for each allele with at least 100 measurements in the training set (118 alleles). The models were trained on a random 80% sample of the data for the allele and the remaining 20% was used for early stopping. All models use the same architecture. The predictions are taken to be the geometric mean of the nM binding affinity predictions of the individual models. The training script is here.

Environment variables

The path where MHCflurry looks for model weights and data can be set with the MHCFLURRY_DOWNLOADS_DIR environment variable. This directory should contain subdirectories like “models_class1”.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

mhcflurry-0.9.1.tar.gz (43.4 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page