search

openvax / mhcflurry

Peptide-MHC I binding affinity prediction
http://openvax.github.io/mhcflurry/
Apache License 2.0
193 stars 58 forks source link

Getting different results than README #106

Closed mhalagan closed 7 years ago

mhalagan commented 7 years ago

From README.md:

>>> 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

When running locally:

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

  allele   peptide   prediction  prediction_low  prediction_high
  A0201  SIINFEKL  5326.547852     3757.866751      7461.377638
timodonnell commented 7 years ago

Just updated the readme with numbers from 0.9.2 release. In general the numerics may differ slightly (up to about 1 nM) even if you match the MHCflurry version due to different Keras backend configurations and other details. Please reopen if you still find discrepancy.