From 492aa09d7dccc2cc579ee33725cfd59c789a6402 Mon Sep 17 00:00:00 2001
From: Tim O'Donnell <timodonnell@gmail.com>
Date: Tue, 20 Mar 2018 16:39:33 -0400
Subject: [PATCH] Fix bug that led to NaN mhcflurry_prediction_high and
mhcflurry_prediction_low uncertainty estimates in some cases
---
mhcflurry/class1_affinity_predictor.py | 4 ++--
test/test_class1_affinity_predictor.py | 8 ++++++++
test/test_predict_command.py | 1 +
3 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/mhcflurry/class1_affinity_predictor.py b/mhcflurry/class1_affinity_predictor.py
index 63cf7960..85ecc1a1 100644
--- a/mhcflurry/class1_affinity_predictor.py
+++ b/mhcflurry/class1_affinity_predictor.py
@@ -1005,8 +1005,8 @@ class Class1AffinityPredictor(object):
df["prediction"] = numpy.exp(log_centers)
if include_confidence_intervals:
- df["prediction_low"] = numpy.exp(numpy.percentile(logs, 5.0, axis=1))
- df["prediction_high"] = numpy.exp(numpy.percentile(logs, 95.0, axis=1))
+ df["prediction_low"] = numpy.exp(numpy.nanpercentile(logs, 5.0, axis=1))
+ df["prediction_high"] = numpy.exp(numpy.nanpercentile(logs, 95.0, axis=1))
if include_individual_model_predictions:
for i in range(num_pan_models):
diff --git a/test/test_class1_affinity_predictor.py b/test/test_class1_affinity_predictor.py
index 30faaf43..3df548a8 100644
--- a/test/test_class1_affinity_predictor.py
+++ b/test/test_class1_affinity_predictor.py
@@ -218,6 +218,14 @@ def test_class1_affinity_predictor_a0205_memorize_training_data():
assert numpy.isnan(ic50_pred[2])
+def test_no_nans():
+ df = DOWNLOADED_PREDICTOR.predict_to_dataframe(
+ alleles=["A02:01", "A02:02"],
+ peptides=["SIINFEKL", "SIINFEKLL"])
+ print(df)
+ assert not df.isnull().any().any()
+
+
def test_predict_implementations_equivalent():
for allele in ["HLA-A02:01", "A02:02"]:
for centrality_measure in ["mean", "robust_mean"]:
diff --git a/test/test_predict_command.py b/test/test_predict_command.py
index cfaf2a0a..f4631020 100644
--- a/test/test_predict_command.py
+++ b/test/test_predict_command.py
@@ -28,6 +28,7 @@ def test_csv():
predict_command.run(full_args)
result = pandas.read_csv(fd_out.name)
print(result)
+ assert not result.isnull().any().any()
finally:
for delete in deletes:
os.unlink(delete)
--
GitLab