Update models_class1_allele_specific_single_kim2014_only to use...

Update models_class1_allele_specific_single_kim2014_only to use hyperparameters roughly the same as preprint

downloads-generation/models_class1_allele_specific_single_kim2014_only/GENERATE.sh

@@ -40,7 +40,7 @@ time mhcflurry-class1-allele-specific-cv-and-train \
     --imputer-description imputer.json \
     --train-data "$(mhcflurry-downloads path data_kim2014)/bdata.2009.mhci.public.1.txt" \
     --test-data "$(mhcflurry-downloads path data_kim2014)/bdata.2013.mhci.public.blind.1.txt" \
-    --min-samples-per-allele 70 \
+    --min-samples-per-allele 50 \
     --out-cv-results cv.csv \
     --out-production-results production.csv \
     --out-models models \

downloads-generation/models_class1_allele_specific_single_kim2014_only/imputer.json

@@ -3,6 +3,6 @@
     "n_burn_in": 5,
     "n_imputations": 50,
     "n_nearest_columns": 25,
-    "min_observations_per_peptide": 5,
-    "min_observations_per_allele": 100 
+    "min_observations_per_peptide": 2,
+    "min_observations_per_allele": 2 
 }

downloads-generation/models_class1_allele_specific_single_kim2014_only/models.py

@@ -3,13 +3,13 @@ from mhcflurry.class1_allele_specific.train import HYPERPARAMETER_DEFAULTS
 import json
 
 models = HYPERPARAMETER_DEFAULTS.models_grid(
-    #impute=[False, True],
-    impute=[False],
+    impute=[False, True],
     activation=["tanh"],
     layer_sizes=[[12], [64], [128]],
     embedding_output_dim=[8, 32, 64],
     dropout_probability=[0, .1, .25],
-    # fraction_negative=[0, .1, .2],
+    pretrain_decay=["1 / (1+epoch)**2"],
+    fraction_negative=[0, .1, .2],
     n_training_epochs=[250])
 
 sys.stderr.write("Models: %d\n" % len(models))