From 1fcf39fc73702cd4a66b063ccf2282de5a3d8a65 Mon Sep 17 00:00:00 2001
From: Tim O'Donnell <timodonnell@gmail.com>
Date: Thu, 3 Oct 2019 16:57:48 -0400
Subject: [PATCH] fixes
---
.../data_mass_spec_benchmark/GENERATE.sh | 25 +++++++++++++++++--
.../run_predictors.py | 21 +++++++++++-----
2 files changed, 38 insertions(+), 8 deletions(-)
diff --git a/downloads-generation/data_mass_spec_benchmark/GENERATE.sh b/downloads-generation/data_mass_spec_benchmark/GENERATE.sh
index bab80ffe..d67bb086 100755
--- a/downloads-generation/data_mass_spec_benchmark/GENERATE.sh
+++ b/downloads-generation/data_mass_spec_benchmark/GENERATE.sh
@@ -1,6 +1,27 @@
#!/bin/bash
#
-# GENERATE.sh <local|cluster> <reuse-all|reuse-none|reuse-predictions>
+# This download includes predictions for MHCflurry and NetMHCpan 4.0 over a
+# large number of peptides encompassing almost the full proteome.
+#
+# Usage:
+# GENERATE.sh <local|cluster> <reuse-all|reuse-none|reuse-predictions|reuse-predictions-except-mhcflurry>
+#
+# The first choice listed above for each argument is the default.
+#
+# Meanings for these arguments:
+#
+# FIRST ARGUMENT: where to run
+# local - run locally using NUM_JOBS cores.
+# cluster - run on cluster.
+#
+# SECOND ARGUMENT: whether to reuse predictions from existing downloaded data
+# reuse-all - reuse predictions and peptide / allele lists from existing
+# downloaded data_mass_spec_benchmark.
+# reuse-none - fully self-contained run; do not reuse anything.
+# reuse-predictions - reuse predictions but not peptide or allele lists. Any
+# new peptides not already included will be run.
+# reuse-predictions-except-mhcflurry
+# - Reuse predictions except for mhcflurry.
#
set -e
set -x
@@ -105,7 +126,7 @@ do
then
REUSE_ARG="--reuse-predictions predictions/chr1.mhcflurry.${kind}"
fi
- if [ "${2:-reuse-none}" != "reuse-none" ]
+ if [ "${2:-reuse-none}" != "reuse-none" ] && [ "${2:-reuse-none}" != "reuse-predictions-except-mhcflurry" ]
then
REUSE_ARG+="--reuse-predictions" "$EXISTING_DATA/$OUT_DIR"
fi
diff --git a/downloads-generation/data_mass_spec_benchmark/run_predictors.py b/downloads-generation/data_mass_spec_benchmark/run_predictors.py
index a86ce0da..ab1d6d1a 100644
--- a/downloads-generation/data_mass_spec_benchmark/run_predictors.py
+++ b/downloads-generation/data_mass_spec_benchmark/run_predictors.py
@@ -81,6 +81,11 @@ parser.add_argument(
add_local_parallelism_args(parser)
add_cluster_parallelism_args(parser)
+PREDICTOR_TO_COLS = {
+ "mhcflurry": ["affinity"],
+ "netmhcpan4": ["affinity", "percentile_rank", "elution_score"],
+}
+
def load_results(dirname, result_df=None):
peptides = pandas.read_csv(
@@ -188,6 +193,7 @@ def run(argv=sys.argv[1:]):
GLOBAL_DATA["predictor"] = args.predictor
GLOBAL_DATA["args"] = args
+ GLOBAL_DATA["cols"] = PREDICTOR_TO_COLS[args.predictor]
# Write peptide and allele lists to out dir.
out_peptides = os.path.abspath(os.path.join(args.out, "peptides.csv"))
@@ -196,7 +202,7 @@ def run(argv=sys.argv[1:]):
manifest_df = []
for allele in alleles:
- for col in ["affinity", "percentile_rank", "elution_score"]:
+ for col in PREDICTOR_TO_COLS[args.predictor]:
manifest_df.append((allele, col))
manifest_df = pandas.DataFrame(
manifest_df, columns=["allele", "kind"])
@@ -222,7 +228,7 @@ def run(argv=sys.argv[1:]):
result_df.notnull().values.mean()))
# We rerun any alleles have nulls for any kind of values
- # (affinity, percentile rank, elution score).
+ # (e.g. affinity, percentile rank, elution score).
print("Computing blocks.")
start = time.time()
blocks = blocks_of_ones(result_df.isnull().values)
@@ -327,7 +333,8 @@ def run(argv=sys.argv[1:]):
prediction_time / 60.0))
-def do_predictions_mhctools(work_item_num, peptides, alleles, constant_data=None):
+def do_predictions_mhctools(
+ work_item_num, peptides, alleles, constant_data=None):
# This may run on the cluster in a way that misses all top level imports,
# so we have to re-import everything here.
import time
@@ -345,6 +352,8 @@ def do_predictions_mhctools(work_item_num, peptides, alleles, constant_data=None
else:
raise ValueError("Unsupported", predictor_name)
+ cols = constant_data['cols']
+
start = time.time()
df = predictor.predict_peptides_dataframe(peptides)
print("Generated predictions for %d peptides x %d alleles in %0.2f sec." % (
@@ -352,9 +361,8 @@ def do_predictions_mhctools(work_item_num, peptides, alleles, constant_data=None
results = {}
for (allele, sub_df) in df.groupby("allele"):
- for col in ["affinity", "percentile_rank", "elution_score"]:
- results["%s %s" % (allele, col)] = sub_df[col].values.astype(
- 'float32')
+ for col in cols:
+ results["%s %s" % (allele, col)] = sub_df[col].values.astype('float32')
return (work_item_num, results)
@@ -371,6 +379,7 @@ def do_predictions_mhcflurry(work_item_num, peptides, alleles, constant_data=Non
args = constant_data['args']
assert args.predictor == "mhcflurry"
+ assert constant_data['cols'] == ["affinity"]
predictor = Class1AffinityPredictor.load(args.mhcflurry_models_dir)
--
GitLab