diff --git a/mhcflurry/__init__.py b/mhcflurry/__init__.py
index 8538c4d3a9afb4f3b3b07b4de7c38c85f765be0a..5d7ceb0fedefd35efb370b702acaf6df0bf799cc 100644
--- a/mhcflurry/__init__.py
+++ b/mhcflurry/__init__.py
@@ -1,3 +1,7 @@
+"""
+Class I MHC ligand prediction package
+"""
+
from .class1_affinity_predictor import Class1AffinityPredictor
from .class1_neural_network import Class1NeuralNetwork
from .version import __version__
diff --git a/mhcflurry/allele_encoding.py b/mhcflurry/allele_encoding.py
index 2e89bfaf2482508d0715becf37328d3e9bc34d52..06355361dd337c5faaa5aa040932a62a733ece87 100644
--- a/mhcflurry/allele_encoding.py
+++ b/mhcflurry/allele_encoding.py
@@ -138,4 +138,3 @@ class AlleleEncoding(object):
result = vector_encoded[self.indices]
self.encoding_cache[cache_key] = result
return self.encoding_cache[cache_key]
-
diff --git a/mhcflurry/calibrate_percentile_ranks_command.py b/mhcflurry/calibrate_percentile_ranks_command.py
index 31aeccc190c6f411c8fedb5d1610ab347f03c61c..9ced4161ea4617a132ec94f677e5c92743ba04fc 100644
--- a/mhcflurry/calibrate_percentile_ranks_command.py
+++ b/mhcflurry/calibrate_percentile_ranks_command.py
@@ -212,7 +212,7 @@ def run(argv=sys.argv[1:]):
worker_pool.join()
print("Percent rank calibration time: %0.2f min." % (
- percent_rank_calibration_time / 60.0))
+ percent_rank_calibration_time / 60.0))
print("Predictor written to: %s" % args.models_dir)
diff --git a/mhcflurry/class1_affinity_predictor.py b/mhcflurry/class1_affinity_predictor.py
index eca348898af74afd5780adc29ade0f929a1e39c6..10ae6365f29829b14fd2f9f7858027c4ca3bfaf3 100644
--- a/mhcflurry/class1_affinity_predictor.py
+++ b/mhcflurry/class1_affinity_predictor.py
@@ -9,12 +9,13 @@ from os import mkdir, environ
from socket import gethostname
from getpass import getuser
from functools import partial
+from six import string_types
-import mhcnames
import numpy
-import pandas
from numpy.testing import assert_equal
-from six import string_types
+import pandas
+
+import mhcnames
from .class1_neural_network import Class1NeuralNetwork
from .common import random_peptides, positional_frequency_matrix
@@ -57,11 +58,11 @@ class Class1AffinityPredictor(object):
Parameters
----------
allele_to_allele_specific_models : dict of string -> list of `Class1NeuralNetwork`
- Ensemble of single-allele models to use for each allele.
-
+ Ensemble of single-allele models to use for each allele.
+
class1_pan_allele_models : list of `Class1NeuralNetwork`
Ensemble of pan-allele models.
-
+
allele_to_sequence : dict of string -> string
MHC allele name to fixed-length amino acid sequence (sometimes
referred to as the pseudosequence). Required only if
@@ -106,7 +107,7 @@ class Class1AffinityPredictor(object):
self._cache = {}
self.optimization_info = {}
- assert isinstance( self.allele_to_allele_specific_models, dict)
+ assert isinstance(self.allele_to_allele_specific_models, dict)
assert isinstance(self.class1_pan_allele_models, list)
@property
@@ -365,14 +366,14 @@ class Class1AffinityPredictor(object):
weights_path = self.weights_path(models_dir, row.model_name)
Class1AffinityPredictor.save_weights(
row.model.get_weights(), weights_path)
- logging.info("Wrote: %s" % weights_path)
+ logging.info("Wrote: %s", weights_path)
write_manifest_df = self.manifest_df[[
c for c in self.manifest_df.columns if c != "model"
]]
manifest_path = join(models_dir, "manifest.csv")
write_manifest_df.to_csv(manifest_path, index=False)
- logging.info("Wrote: %s" % manifest_path)
+ logging.info("Wrote: %s", manifest_path)
if write_metadata:
# Write "info.txt"
@@ -399,7 +400,7 @@ class Class1AffinityPredictor(object):
)
allele_to_sequence_df.to_csv(
join(models_dir, "allele_sequences.csv"), index=False)
- logging.info("Wrote: %s" % join(models_dir, "allele_sequences.csv"))
+ logging.info("Wrote: %s", join(models_dir, "allele_sequences.csv"))
if self.allele_to_percent_rank_transform:
percent_ranks_df = None
@@ -414,7 +415,7 @@ class Class1AffinityPredictor(object):
percent_ranks_path,
index=True,
index_label="bin")
- logging.info("Wrote: %s" % percent_ranks_path)
+ logging.info("Wrote: %s", percent_ranks_path)
@staticmethod
def load(models_dir=None, max_models=None):
@@ -467,7 +468,7 @@ class Class1AffinityPredictor(object):
if exists(join(models_dir, "allele_sequences.csv")):
allele_to_sequence = pandas.read_csv(
join(models_dir, "allele_sequences.csv"),
- index_col=0).iloc[:,0].to_dict()
+ index_col=0).iloc[:, 0].to_dict()
allele_to_percent_rank_transform = {}
percent_ranks_path = join(models_dir, "percent_ranks.csv")
@@ -479,15 +480,15 @@ class Class1AffinityPredictor(object):
logging.info(
"Loaded %d class1 pan allele predictors, %d allele sequences, "
- "%d percent rank distributions, and %d allele specific models: %s" % (
- len(class1_pan_allele_models),
- len(allele_to_sequence) if allele_to_sequence else 0,
- len(allele_to_percent_rank_transform),
- sum(len(v) for v in allele_to_allele_specific_models.values()),
- ", ".join(
- "%s (%d)" % (allele, len(v))
- for (allele, v)
- in sorted(allele_to_allele_specific_models.items()))))
+ "%d percent rank distributions, and %d allele specific models: %s",
+ len(class1_pan_allele_models),
+ len(allele_to_sequence) if allele_to_sequence else 0,
+ len(allele_to_percent_rank_transform),
+ sum(len(v) for v in allele_to_allele_specific_models.values()),
+ ", ".join(
+ "%s (%d)" % (allele, len(v))
+ for (allele, v)
+ in sorted(allele_to_allele_specific_models.items())))
result = Class1AffinityPredictor(
allele_to_allele_specific_models=allele_to_allele_specific_models,
@@ -500,7 +501,7 @@ class Class1AffinityPredictor(object):
logging.info("Optimizing models")
optimized = result.optimize()
logging.info(
- "Optimization " + ("succeeded" if optimized else "failed"))
+ "Optimization %s", ("succeeded" if optimized else "failed"))
return result
def optimize(self):
@@ -527,7 +528,7 @@ class Class1AffinityPredictor(object):
merge_method="concatenate")
]
except NotImplementedError as e:
- logging.warning("Optimization failed: %s" % str(e))
+ logging.warning("Optimization failed: %s", str(e))
return False
self._manifest_df = None
self.clear_cache()
@@ -584,8 +585,8 @@ class Class1AffinityPredictor(object):
AlleleEncoding
"""
if (self._master_allele_encoding is None or
- self._master_allele_encoding.allele_to_sequence !=
- self.allele_to_sequence):
+ self._master_allele_encoding.allele_to_sequence !=
+ self.allele_to_sequence):
self._master_allele_encoding = AlleleEncoding(
allele_to_sequence=self.allele_to_sequence)
return self._master_allele_encoding
@@ -793,7 +794,7 @@ class Class1AffinityPredictor(object):
encodable_peptides = EncodableSequences.create(peptides)
models = []
for i in range(n_models):
- logging.info("Training model %d / %d" % (i + 1, n_models))
+ logging.info("Training model %d / %d", i + 1, n_models)
model = Class1NeuralNetwork(**architecture_hyperparameters)
model.fit(
encodable_peptides,
@@ -879,10 +880,8 @@ class Class1AffinityPredictor(object):
msg = "Allele %s has no percentile rank information" % allele
if throw:
raise ValueError(msg)
- else:
- warnings.warn(msg)
- # Return NaNs
- return numpy.ones(len(affinities)) * numpy.nan
+ warnings.warn(msg)
+ return numpy.ones(len(affinities)) * numpy.nan # Return NaNs
if alleles is None:
raise ValueError("Specify allele or alleles")
@@ -1294,10 +1293,10 @@ class Class1AffinityPredictor(object):
Returns
----------
- None if motif_summary is False
+ dict of string -> pandas.DataFrame
- Otherwise: dict of string -> pandas.DataFrame where keys are
- "frequency_matrices" and "length_distributions".
+ If motif_summary is True, this will have keys "frequency_matrices" and
+ "length_distributions". Otherwise it will be empty.
"""
if bins is None:
@@ -1323,7 +1322,7 @@ class Class1AffinityPredictor(object):
else:
frequency_matrices = None
length_distributions = None
- for (i, allele) in enumerate(alleles):
+ for allele in alleles:
start = time.time()
predictions = self.predict(
encoded_peptides, allele=allele, model_kwargs=model_kwargs)
@@ -1400,6 +1399,7 @@ class Class1AffinityPredictor(object):
'frequency_matrices': frequency_matrices,
'length_distributions': length_distributions,
}
+ return {}
def model_select(
self,
@@ -1490,4 +1490,3 @@ class Class1AffinityPredictor(object):
"model_selection": df,
})
return new_predictor
-
diff --git a/mhcflurry/cluster_parallelism.py b/mhcflurry/cluster_parallelism.py
index 976f53a213b20c261bd4d9cf288430652694a0d5..9ec07de96283da2a351053297fee719f7cb20499 100644
--- a/mhcflurry/cluster_parallelism.py
+++ b/mhcflurry/cluster_parallelism.py
@@ -349,4 +349,3 @@ def worker_entry_point(argv=sys.argv[1:]):
if args.complete_dir:
os.mkdir(args.complete_dir)
print("Created: ", args.complete_dir)
-
diff --git a/mhcflurry/common.py b/mhcflurry/common.py
index 7c8e1628682f1d716d2b3dc866c85eb2b897e176..8885637b2e963c98b70dcadf8762cd43cbc2ff4b 100644
--- a/mhcflurry/common.py
+++ b/mhcflurry/common.py
@@ -173,4 +173,4 @@ def positional_frequency_matrix(peptides):
counts[i + 1] = pandas.Series([p[i] for p in peptides]).value_counts()
result = (counts / len(peptides)).fillna(0.0).T
result.index.name = 'position'
- return result
\ No newline at end of file
+ return result
diff --git a/mhcflurry/custom_loss.py b/mhcflurry/custom_loss.py
index 47d51cb49aa5ffccbcb4dc714d97ba07e1289e91..8c523c0a61ad0576d54c1cd77a15dfdf9ba987c8 100644
--- a/mhcflurry/custom_loss.py
+++ b/mhcflurry/custom_loss.py
@@ -251,4 +251,4 @@ def check_shape(name, arr, expected_shape):
# Register custom losses.
for cls in [MSEWithInequalities, MSEWithInequalitiesAndMultipleOutputs]:
- CUSTOM_LOSSES[cls.name] = cls()
\ No newline at end of file
+ CUSTOM_LOSSES[cls.name] = cls()
diff --git a/mhcflurry/downloads.py b/mhcflurry/downloads.py
index ec776f9e093c7309e5f06a54d0b35a52489d0d96..7f29ea521cb558cbf47523b1302b2875041509a5 100644
--- a/mhcflurry/downloads.py
+++ b/mhcflurry/downloads.py
@@ -9,9 +9,9 @@ from __future__ import (
)
import logging
import yaml
-from os.path import join, exists, relpath
-from pipes import quote
+from os.path import join, exists
from os import environ
+from pipes import quote
from collections import OrderedDict
from appdirs import user_data_dir
from pkg_resources import resource_string
@@ -81,8 +81,7 @@ def get_default_class1_models_dir(test_exists=True):
if test_exists and not exists(result):
raise IOError("No such directory: %s" % result)
return result
- else:
- return get_path("models_class1", "models", test_exists=test_exists)
+ return get_path("models_class1", "models", test_exists=test_exists)
def get_current_release_downloads():
@@ -160,13 +159,13 @@ def configure():
metadata["releases"][_CURRENT_RELEASE]["compatibility-version"])
current_compatability = metadata["current-compatibility-version"]
if current_release_compatability != current_compatability:
- logging.warn(
+ logging.warning(
"The specified downloads are not compatible with this version "
"of the MHCflurry codebase. Downloads: release %s, "
- "compatability version: %d. Code compatability version: %d" % (
- _CURRENT_RELEASE,
- current_release_compatability,
- current_compatability))
+ "compatability version: %d. Code compatability version: %d",
+ _CURRENT_RELEASE,
+ current_release_compatability,
+ current_compatability)
data_dir = environ.get("MHCFLURRY_DATA_DIR")
if not data_dir:
@@ -176,6 +175,7 @@ def configure():
data_dir = user_data_dir("mhcflurry", version="4")
_DOWNLOADS_DIR = join(data_dir, _CURRENT_RELEASE)
- logging.debug("Configured MHCFLURRY_DOWNLOADS_DIR: %s" % _DOWNLOADS_DIR)
+ logging.debug("Configured MHCFLURRY_DOWNLOADS_DIR: %s", _DOWNLOADS_DIR)
+
configure()
diff --git a/mhcflurry/downloads_command.py b/mhcflurry/downloads_command.py
index 8bf1d213400ddc3c2a7e5d9b6e81ae689189f54e..249269aea94f3a8b3bdcec5f7fce26f41c4e804d 100644
--- a/mhcflurry/downloads_command.py
+++ b/mhcflurry/downloads_command.py
@@ -168,8 +168,7 @@ def fetch_subcommand(args):
"\nThe requested download '%s' has already been downloaded. "
"To re-download this data, first run: \n\t%s\nin a shell "
"and then re-run this command.\n" +
- "*" * 40)
- % (name, 'rm -rf ' + quote(get_path(name))))
+ "*" * 40) % (name, 'rm -rf ' + quote(get_path(name))))
if not info['downloaded'] and (name in args.download_name or default):
items_to_fetch.add(name)
diff --git a/mhcflurry/encodable_sequences.py b/mhcflurry/encodable_sequences.py
index f6322835c92af973d673336c4a969e74f3182597..19696e23b9f901811e0fb01b9f16f58a7c0ac613 100644
--- a/mhcflurry/encodable_sequences.py
+++ b/mhcflurry/encodable_sequences.py
@@ -1,3 +1,6 @@
+"""
+Class for encoding variable-length peptides to fixed-size numerical matrices
+"""
from __future__ import (
print_function,
division,
@@ -26,9 +29,12 @@ class EncodingError(ValueError):
class EncodableSequences(object):
"""
- Sequences of amino acids.
+ Class for encoding variable-length peptides to fixed-size numerical matrices
This class caches various encodings of a list of sequences.
+
+ In practice this is used only for peptides. To encode MHC allele sequences,
+ see AlleleEncoding.
"""
unknown_character = "X"
@@ -299,8 +305,10 @@ class EncodableSequences(object):
min_length = 5
# Result array is int32, filled with X (null amino acid) value.
- result = numpy.full(fill_value=amino_acid.AMINO_ACID_INDEX['X'],
- shape=(len(sequences), max_length * 2), dtype="int32")
+ result = numpy.full(
+ fill_value=amino_acid.AMINO_ACID_INDEX['X'],
+ shape=(len(sequences), max_length * 2),
+ dtype="int32")
df = pandas.DataFrame({"peptide": sequences}, dtype=numpy.object_)
@@ -319,9 +327,9 @@ class EncodableSequences(object):
# Array of shape (num peptides, length) giving fixed-length
# amino acid encoding each peptide of the current length.
fixed_length_sequences = numpy.stack(sub_df.peptide.map(
- lambda s: numpy.array(
- [amino_acid.AMINO_ACID_INDEX[char] for char in
- s])).values)
+ lambda s: numpy.array([
+ amino_acid.AMINO_ACID_INDEX[char] for char in s
+ ])).values)
# Set left edge
result[sub_df.index, :length] = fixed_length_sequences
@@ -334,8 +342,10 @@ class EncodableSequences(object):
min_length = 5
# Result array is int32, filled with X (null amino acid) value.
- result = numpy.full(fill_value=amino_acid.AMINO_ACID_INDEX['X'],
- shape=(len(sequences), max_length * 3), dtype="int32")
+ result = numpy.full(
+ fill_value=amino_acid.AMINO_ACID_INDEX['X'],
+ shape=(len(sequences), max_length * 3),
+ dtype="int32")
df = pandas.DataFrame({"peptide": sequences}, dtype=numpy.object_)
@@ -354,9 +364,9 @@ class EncodableSequences(object):
# Array of shape (num peptides, length) giving fixed-length
# amino acid encoding each peptide of the current length.
fixed_length_sequences = numpy.stack(sub_df.peptide.map(
- lambda s: numpy.array(
- [amino_acid.AMINO_ACID_INDEX[char] for char in
- s])).values)
+ lambda s: numpy.array([
+ amino_acid.AMINO_ACID_INDEX[char] for char in s
+ ])).values)
# Set left edge
result[sub_df.index, :length] = fixed_length_sequences
diff --git a/mhcflurry/ensemble_centrality.py b/mhcflurry/ensemble_centrality.py
index e370a39d66f31d8e343605a22c38e62bd12160b0..07251bf028e5414abf31afc53db2b2a2abb96e53 100644
--- a/mhcflurry/ensemble_centrality.py
+++ b/mhcflurry/ensemble_centrality.py
@@ -37,4 +37,4 @@ CENTRALITY_MEASURES = {
"mean": partial(numpy.nanmean, axis=1),
"median": partial(numpy.nanmedian, axis=1),
"robust_mean": robust_mean,
-}
\ No newline at end of file
+}
diff --git a/mhcflurry/hyperparameters.py b/mhcflurry/hyperparameters.py
index cc5950d5c175a35c08eb9cfea0d44d64e3fdeb65..1241fa4650378f17247d2c0bea529ba4ac0a0261 100644
--- a/mhcflurry/hyperparameters.py
+++ b/mhcflurry/hyperparameters.py
@@ -1,3 +1,6 @@
+"""
+Hyperparameter (neural network options) management
+"""
from __future__ import (
print_function,
division,
@@ -70,8 +73,7 @@ class HyperparameterDefaults(object):
if invalid_keys:
raise ValueError(
"No such model parameters: %s. Valid parameters are: %s"
- % (" ".join(invalid_keys),
- " ".join(self.defaults)))
+ % (" ".join(invalid_keys), " ".join(self.defaults)))
def models_grid(self, **kwargs):
'''
diff --git a/mhcflurry/percent_rank_transform.py b/mhcflurry/percent_rank_transform.py
index a9098bc231675ddbe7444066a506f85ae3107660..a4597686941b97162ac3bcf6f833b3856f354e44 100644
--- a/mhcflurry/percent_rank_transform.py
+++ b/mhcflurry/percent_rank_transform.py
@@ -1,3 +1,6 @@
+"""
+Class for transforming arbitrary values into percent ranks given a distribution.
+"""
import numpy
import pandas
@@ -77,8 +80,3 @@ class PercentRankTransform(object):
result.cdf = series.values
result.bin_edges = series.index.values[1:-1]
return result
-
-
-
-
-
diff --git a/mhcflurry/predict_command.py b/mhcflurry/predict_command.py
index dea1ddf4c15cc7ec8c1ea605985c9cbb351631d4..17a687f55eff01b46573c84341088ec9d3ecea88 100644
--- a/mhcflurry/predict_command.py
+++ b/mhcflurry/predict_command.py
@@ -219,7 +219,7 @@ def run(argv=sys.argv[1:]):
})
logging.info(
"Predicting for %d alleles and %d peptides = %d predictions" % (
- len(args.alleles), len(args.peptides), len(df)))
+ len(args.alleles), len(args.peptides), len(df)))
predictions = predictor.predict_to_dataframe(
peptides=df[args.peptide_column].values,
diff --git a/mhcflurry/scoring.py b/mhcflurry/scoring.py
index f6a256baa18e1bbe01968ce270466791588a34f1..d0d41d4e01e215c8c1e603fd5d1ec09dc08cf673 100644
--- a/mhcflurry/scoring.py
+++ b/mhcflurry/scoring.py
@@ -1,3 +1,6 @@
+"""
+Measures of prediction accuracy
+"""
from __future__ import (
print_function,
division,
diff --git a/mhcflurry/select_pan_allele_models_command.py b/mhcflurry/select_pan_allele_models_command.py
index 0032f8fc5b9553e234473c44ca764420d30210e9..6e6f45fa8c29fd5ead83567ce251e9cb88141e08 100644
--- a/mhcflurry/select_pan_allele_models_command.py
+++ b/mhcflurry/select_pan_allele_models_command.py
@@ -24,7 +24,7 @@ tqdm.monitor_interval = 0 # see https://github.com/tqdm/tqdm/issues/481
from .class1_affinity_predictor import Class1AffinityPredictor
from .encodable_sequences import EncodableSequences
from .allele_encoding import AlleleEncoding
-from .common import configure_logging, random_peptides
+from .common import configure_logging
from .local_parallelism import (
worker_pool_with_gpu_assignments_from_args,
add_local_parallelism_args)
diff --git a/mhcflurry/train_allele_specific_models_command.py b/mhcflurry/train_allele_specific_models_command.py
index fe295b67df5c96080f3834c4cf95cfcf9a9d60fe..ec1b8c8af7adac1e10ef6d3305780b86ea98f41a 100644
--- a/mhcflurry/train_allele_specific_models_command.py
+++ b/mhcflurry/train_allele_specific_models_command.py
@@ -10,7 +10,6 @@ import traceback
import random
from functools import partial
-import numpy
import pandas
import yaml
from sklearn.metrics.pairwise import cosine_similarity
@@ -337,7 +336,7 @@ def alleles_by_similarity(allele):
allele_similarity.columns.to_series().sample(frac=1.0))
return (
allele_similarity[allele] + (
- allele_similarity.index == allele) # force that we return specified allele first
+ allele_similarity.index == allele) # force specified allele first
).sort_values(ascending=False).index.tolist()
diff --git a/mhcflurry/train_pan_allele_models_command.py b/mhcflurry/train_pan_allele_models_command.py
index d925f991af0cc4d66cf4bbf081eaa6d0d698ec06..a676bb38f2e64b7c83ae1f8b3bfdfd4ab90bb1b0 100644
--- a/mhcflurry/train_pan_allele_models_command.py
+++ b/mhcflurry/train_pan_allele_models_command.py
@@ -270,7 +270,8 @@ def run(argv=sys.argv[1:]):
return main(args)
except Exception as e:
print(e)
- import ipdb ; ipdb.set_trace()
+ import ipdb
+ ipdb.set_trace()
raise
else:
return main(args)
@@ -697,4 +698,3 @@ def train_model(
if __name__ == '__main__':
run()
-