Closed wwang-nmdp closed 7 years ago
L0504656:mhcflurry wwang$ L0504656:mhcflurry wwang$ mhcflurry-predict --alleles HLA-A0201 HLA-A0301 --peptides SIINFEKL SIINFEKD SIINFEKQ Using Theano backend. Predicting for 2 alleles and 3 peptides = 6 predictions Traceback (most recent call last): File "//anaconda/bin/mhcflurry-predict", line 11, in <module> load_entry_point('mhcflurry', 'console_scripts', 'mhcflurry-predict')() File "/Users/wwang/Documents/Tools/mhcflurry/mhcflurry/predict_command.py", line 142, in run model = class1_allele_specific_loader.from_allele_name(allele) File "/Users/wwang/Documents/Tools/mhcflurry/mhcflurry/class1_allele_specific/load.py", line 125, in from_allele_name self.predictors_cache[allele_name] = pickle.load(fd) File "/Users/wwang/Documents/Tools/mhcflurry/mhcflurry/class1_allele_specific/class1_binding_predictor.py", line 132, in __setstate__ self.model = keras.models.load_model(fd.name) File "//anaconda/lib/python3.5/site-packages/keras/models.py", line 167, in load_model model.optimizer.set_weights(optimizer_weight_values) File "//anaconda/lib/python3.5/site-packages/keras/optimizers.py", line 97, in set_weights 'provided weight shape ' + str(w.shape)) Exception: Optimizer weight shape (72, 64) not compatible with provided weight shape (64,) L0504656:mhcflurry wwang$
My hunch is this is due to keras changing constantly. We should probably pin the keras version.
Confirmed broken by re-running travis build on master: https://travis-ci.org/hammerlab/mhcflurry/jobs/162961786