def test_var_eff_pred_varseq(tmpdir):
model_name = "DeepSEA/variantEffects"
if INSTALL_REQ:
install_model_requirements(model_name, "kipoi", and_dataloaders=True)
#
model = kipoi.get_model(model_name, source="kipoi")
# The preprocessor
Dataloader = SeqIntervalDl
#
dataloader_arguments = {"intervals_file": "example_files/intervals.bed",
"fasta_file": "example_files/hg38_chr22.fa",
"required_seq_len": 1000, "alphabet_axis": 1, "dummy_axis": 2, "label_dtype": str}
dataloader_arguments = {k: model.source_dir + "/" + v if isinstance(v, str) else v for k, v in
dataloader_arguments.items()}
vcf_path = "tests/data/variants.vcf"
out_vcf_fpath = str(tmpdir.mkdir("variants_generated", ).join("out.vcf"))
#
vcf_path = kipoi_veff.ensure_tabixed_vcf(vcf_path)
model_info = kipoi_veff.ModelInfoExtractor(model, Dataloader)
writer = kipoi_veff.VcfWriter(
model, vcf_path, out_vcf_fpath, standardise_var_id=True)
vcf_to_region = kipoi_veff.SnvCenteredRg(model_info)
res = sp.predict_snvs(model, Dataloader, vcf_path, dataloader_args=dataloader_arguments,
batch_size=32,
vcf_to_region=vcf_to_region,
sync_pred_writer=writer)
writer.close()
assert os.path.exists(out_vcf_fpath)
def add_scores(self, snp_vcf_path='../data/snp_vcfs', out_dir=None):
if out_dir is None:
out_dir = '../data/model_scores/' + self.model_name
if not os.path.exists(out_dir):
os.makedirs(out_dir)
file_names = os.listdir(self.snp_vcf_path)
for file_name in file_names:
chrom = file_name.split('.')[0]
Dataloader = self.model.default_dataloader
vcf_path = self.snp_vcf_path + '/' + file_name
out_vcf_fpath = out_dir + '/' + chrom + '.vcf'
print(vcf_path)
print(out_vcf_fpath)
writer = VcfWriter(self.model, vcf_path, out_vcf_fpath)
model_info = kipoi_veff.ModelInfoExtractor(self.model, Dataloader)
# vcf_to_region will generate a variant-centered regions when presented a VCF record.
vcf_to_region = kipoi_veff.SnvCenteredRg(model_info)
dataloader_arguments = {
"fasta_file": '../data/fasta_files/chr' + chrom + '.fa'
}
sp.predict_snvs(
self.model,
Dataloader,
vcf_path,
batch_size=32,
dataloader_args=dataloader_arguments,
vcf_to_region=vcf_to_region,
#evaluation_function_kwargs={'diff_types': {'diff': Diff("mean"), 'deepsea_effect': DeepSEA_effect("mean")}},
sync_pred_writer=writer)
def test_mutation_map():
if sys.version_info[0] == 2:
pytest.skip("rbp example not supported on python 2 ")
# Take the rbp model
model_dir = "tests/models/rbp/"
if INSTALL_REQ:
install_model_requirements(model_dir, "dir", and_dataloaders=True)
model = kipoi.get_model(model_dir, source="dir")
# The preprocessor
Dataloader = kipoi.get_dataloader_factory(model_dir, source="dir")
#
dataloader_arguments = {
"fasta_file": "example_files/hg38_chr22.fa",
"preproc_transformer": "dataloader_files/encodeSplines.pkl",
"gtf_file": "example_files/gencode_v25_chr22.gtf.pkl.gz",
}
dataloader_arguments = {
k: model_dir + v
for k, v in dataloader_arguments.items()
}
#
# Run the actual predictions
vcf_path = model_dir + "example_files/first_variant.vcf"
#
model_info = kipoi_veff.ModelInfoExtractor(model, Dataloader)
vcf_to_region = kipoi_veff.SnvCenteredRg(model_info)
mdmm = mm._generate_mutation_map(
model,
Dataloader,
vcf_path,
dataloader_args=dataloader_arguments,
evaluation_function=analyse_model_preds,
batch_size=32,
vcf_to_region=vcf_to_region,
evaluation_function_kwargs={'diff_types': {
'diff': Diff("mean")
}})
with cd(model.source_dir):
mdmm.save_to_file("example_files/first_variant_mm_totest.hdf5")
from kipoi_veff.utils.generic import read_hdf5
reference = read_hdf5("example_files/first_variant_mm.hdf5")
obs = read_hdf5("example_files/first_variant_mm.hdf5")
compare_rec(reference[0], obs[0])
import matplotlib
matplotlib.pyplot.switch_backend('agg')
mdmm.plot_mutmap(0, "seq", "diff", "rbp_prb")
os.unlink("example_files/first_variant_mm_totest.hdf5")
import pandas as pd
model_name = "DeepBind/Homo_sapiens/TF/D00299.003_SELEX_ATF7"
# get the model
model = kipoi.get_model(model_name)
# get the dataloader factory
Dataloader = model.default_dataloader
vcf_path = "../data/test.vcf"
# The output vcf path, based on the input file name
out_vcf_fpath = vcf_path[:-4] + "%s.vcf" % model_name.replace("/", "_")
# The writer object that will output the annotated VCF
writer = VcfWriter(model, vcf_path, out_vcf_fpath)
# Information extraction from dataloader and model
model_info = kipoi_veff.ModelInfoExtractor(model, Dataloader)
# vcf_to_region will generate a variant-centered regions when presented a VCF record.
vcf_to_region = kipoi_veff.SnvCenteredRg(model_info)
dataloader_arguments = {"fasta_file": "../data/fasta_files/chr1.fa"}
sp.predict_snvs(
model,
Dataloader,
vcf_path,
batch_size=32,
dataloader_args=dataloader_arguments,
vcf_to_region=vcf_to_region,
#evaluation_function_kwargs={'diff_types': {'diff': Diff("mean"), 'deepsea_effect': DeepSEA_effect("mean")}},
sync_pred_writer=writer)
vcf_reader = KipoiVCFParser(out_vcf_fpath)
def cli_score_variants(command, raw_args):
"""CLI interface to score variants
"""
# Updated argument names:
# - scoring -> scores
# - --vcf_path -> --input_vcf, -i
# - --out_vcf_fpath -> --output_vcf, -o
# - --output -> -e, --extra_output
# - remove - -install_req
# - scoring_kwargs -> score_kwargs
AVAILABLE_FORMATS = [k for k in writers.FILE_SUFFIX_MAP if k != 'bed']
assert command == "score_variants"
parser = argparse.ArgumentParser(
'kipoi veff {}'.format(command),
description='Predict effect of SNVs using ISM.')
parser.add_argument('model', help='Model name.')
parser.add_argument(
'--source',
default="kipoi",
choices=list(kipoi.config.model_sources().keys()),
help='Model source to use. Specified in ~/.kipoi/config.yaml' +
" under model_sources. " +
"'dir' is an additional source referring to the local folder.")
add_dataloader(parser=parser, with_args=True)
parser.add_argument('-i', '--input_vcf', required=True, help='Input VCF.')
parser.add_argument('-o',
'--output_vcf',
help='Output annotated VCF file path.',
default=None)
parser.add_argument('--batch_size',
type=int,
default=32,
help='Batch size to use in prediction')
parser.add_argument(
"-n",
"--num_workers",
type=int,
default=0,
help="Number of parallel workers for loading the dataset")
parser.add_argument(
'-r',
'--restriction_bed',
default=None,
help="Regions for prediction can only be subsets of this bed file")
parser.add_argument(
'-e',
'--extra_output',
type=str,
default=None,
required=False,
help=
"Additional output files in other (non-vcf) formats. File format is inferred from the file path ending"
+ ". Available file formats are: {0}".format(", ".join(
["." + k for k in AVAILABLE_FORMATS])))
parser.add_argument(
'-s',
"--scores",
default="diff",
nargs="+",
help=
"Scoring method to be used. Only scoring methods selected in the model yaml file are"
"available except for `diff` which is always available. Select scoring function by the"
"`name` tag defined in the model yaml file.")
parser.add_argument(
'-k',
"--score_kwargs",
default="",
nargs="+",
help=
"JSON definition of the kwargs for the scoring functions selected in --scoring. The "
"definiton can either be in JSON in the command line or the path of a .json file. The "
"individual JSONs are expected to be supplied in the same order as the labels defined in "
"--scoring. If the defaults or no arguments should be used define '{}' for that respective "
"scoring method.")
parser.add_argument(
'-l',
"--seq_length",
type=int,
default=None,
help=
"Optional parameter: Model input sequence length - necessary if the model does not have a "
"pre-defined input sequence length.")
parser.add_argument(
'--std_var_id',
action="store_true",
help="If set then variant IDs in the annotated"
" VCF will be replaced with a standardised, unique ID.")
parser.add_argument(
"--model_outputs",
type=str,
default=None,
nargs="+",
help=
"Optional parameter: Only return predictions for the selected model outputs. Naming"
"according to the definition in model.yaml > schema > targets > column_labels"
)
parser.add_argument(
"--model_outputs_i",
type=int,
default=None,
nargs="+",
help=
"Optional parameter: Only return predictions for the selected model outputs. Give integer"
"indices of the selected model output(s).")
parser.add_argument(
"--singularity",
action='store_true',
help="Run `kipoi predict` in the appropriate singularity container. "
"Containters will get downloaded to ~/.kipoi/envs/ or to "
"$SINGULARITY_CACHEDIR if set")
args = parser.parse_args(raw_args)
# OBSOLETE
# Make sure all the multi-model arguments like source, dataloader etc. fit together
#_prepare_multi_model_args(args)
# Check that all the folders exist
file_exists(args.input_vcf, logger)
if args.output_vcf is None and args.extra_output is None:
logger.error(
"One of the two needs to be specified: --output_vcf or --extra_output"
)
sys.exit(1)
if args.extra_output is not None:
dir_exists(os.path.dirname(args.extra_output), logger)
ending = args.extra_output.split('.')[-1]
if ending not in AVAILABLE_FORMATS:
logger.error("File ending: {0} for file {1} not from {2}".format(
ending, args.extra_output, AVAILABLE_FORMATS))
sys.exit(1)
# singularity_command
if args.singularity:
from kipoi.cli.singularity import singularity_command
logger.info(
"Running kipoi veff {} in the singularity container".format(
command))
# Drop the singularity flag
raw_args = [x for x in raw_args if x != '--singularity']
dataloader_kwargs = parse_json_file_str_or_arglist(
args.dataloader_args)
# create output files
output_files = []
if args.output_vcf is not None:
output_files.append(args.output_vcf)
if args.extra_output is not None:
output_files.append(args.extra_output)
singularity_command(['kipoi', 'veff', command] + raw_args,
model=args.model,
dataloader_kwargs=dataloader_kwargs,
output_files=output_files,
source=args.source,
dry_run=False)
return None
if not isinstance(args.scores, list):
args.scores = [args.scores]
score_kwargs = []
if len(args.score_kwargs) > 0:
score_kwargs = args.score_kwargs
if len(args.scores) >= 1:
# Check if all scoring functions should be used:
if args.scores == ["all"]:
if len(score_kwargs) >= 1:
raise ValueError(
"`--score_kwargs` cannot be defined in combination will `--scoring all`!"
)
else:
score_kwargs = [parse_json_file_str(el) for el in score_kwargs]
if not len(args.score_kwargs) == len(score_kwargs):
raise ValueError(
"When defining `--score_kwargs` a JSON representation of arguments (or the "
"path of a file containing them) must be given for every "
"`--scores` function.")
# VCF writer
output_vcf_model = None
if args.output_vcf is not None:
dir_exists(os.path.dirname(args.output_vcf), logger)
output_vcf_model = args.output_vcf
# Other writers
if args.extra_output is not None:
dir_exists(os.path.dirname(args.extra_output), logger)
extra_output = args.extra_output
writer = writers.get_writer(extra_output, metadata_schema=None)
assert writer is not None
extra_writers = [SyncBatchWriter(writer)]
else:
extra_writers = []
dataloader_arguments = parse_json_file_str_or_arglist(args.dataloader_args)
# --------------------------------------------
# load model & dataloader
model = kipoi.get_model(args.model, args.source)
if args.dataloader is not None:
Dl = kipoi.get_dataloader_factory(args.dataloader,
args.dataloader_source)
else:
Dl = model.default_dataloader
# Load effect prediction related model info
model_info = kipoi_veff.ModelInfoExtractor(model, Dl)
if model_info.use_seq_only_rc:
logger.info(
'Model SUPPORTS simple reverse complementation of input DNA sequences.'
)
else:
logger.info(
'Model DOES NOT support simple reverse complementation of input DNA sequences.'
)
if output_vcf_model is not None:
logger.info('Annotated VCF will be written to %s.' %
str(output_vcf_model))
model_outputs = None
if args.model_outputs is not None:
model_outputs = args.model_outputs
elif args.model_outputs_i is not None:
model_outputs = args.model_outputs_i
kipoi_veff.score_variants(model,
dataloader_arguments,
args.input_vcf,
output_vcf=output_vcf_model,
output_writers=extra_writers,
scores=args.scores,
score_kwargs=score_kwargs,
num_workers=args.num_workers,
batch_size=args.batch_size,
seq_length=args.seq_length,
std_var_id=args.std_var_id,
restriction_bed=args.restriction_bed,
return_predictions=False,
model_outputs=model_outputs)
logger.info('Successfully predicted samples')
def cli_create_mutation_map(command, raw_args):
"""CLI interface to calculate mutation map data
"""
assert command == "create_mutation_map"
parser = argparse.ArgumentParser(
'kipoi veff {}'.format(command),
description='Predict effect of SNVs using ISM.')
add_model(parser)
add_dataloader(parser, with_args=True)
parser.add_argument(
'-r',
'--regions_file',
help='Region definition as VCF or bed file. Not a required input.')
parser.add_argument('--batch_size',
type=int,
default=32,
help='Batch size to use in prediction')
parser.add_argument(
"-n",
"--num_workers",
type=int,
default=0,
help="Number of parallel workers for loading the dataset")
parser.add_argument("-i",
"--install_req",
action='store_true',
help="Install required packages from requirements.txt")
parser.add_argument(
'-o',
'--output',
required=True,
help="Output HDF5 file. To be used as input for plotting.")
parser.add_argument(
'-s',
"--scores",
default="diff",
nargs="+",
help=
"Scoring method to be used. Only scoring methods selected in the model yaml file are"
"available except for `diff` which is always available. Select scoring function by the"
"`name` tag defined in the model yaml file.")
parser.add_argument(
'-k',
"--score_kwargs",
default="",
nargs="+",
help=
"JSON definition of the kwargs for the scoring functions selected in --scores. The "
"definiton can either be in JSON in the command line or the path of a .json file. The "
"individual JSONs are expected to be supplied in the same order as the labels defined in "
"--scores. If the defaults or no arguments should be used define '{}' for that respective "
"scoring method.")
parser.add_argument(
'-l',
"--seq_length",
type=int,
default=None,
help=
"Optional parameter: Model input sequence length - necessary if the model does not have a "
"pre-defined input sequence length.")
parser.add_argument(
"--singularity",
action='store_true',
help="Run `kipoi predict` in the appropriate singularity container. "
"Containters will get downloaded to ~/.kipoi/envs/ or to "
"$SINGULARITY_CACHEDIR if set")
args = parser.parse_args(raw_args)
# extract args for kipoi.variant_effects.predict_snvs
print("DL ARGS", args.dataloader_args)
dataloader_arguments = parse_json_file_str_or_arglist(args.dataloader_args)
#dataloader_arguments = parse_json_file_str(args.dataloader_args)
if args.output is None:
raise Exception("Output file `--output` has to be set!")
if args.singularity:
from kipoi.cli.singularity import singularity_command
logger.info(
"Running kipoi veff in the singularity container".format(command))
# Drop the singularity flag
raw_args = [x for x in raw_args if x != '--singularity']
singularity_command(['kipoi', 'veff', command] + raw_args,
args.model,
dataloader_arguments,
output_files=args.output,
source=args.source,
dry_run=False)
return None
# --------------------------------------------
# install args
if args.install_req:
kipoi.pipeline.install_model_requirements(args.model,
args.source,
and_dataloaders=True)
# load model & dataloader
model = kipoi.get_model(args.model, args.source)
regions_file = os.path.realpath(args.regions_file)
output = os.path.realpath(args.output)
with cd(model.source_dir):
if not os.path.exists(regions_file):
raise Exception("Regions inputs file does not exist: %s" %
args.regions_file)
# Check that all the folders exist
file_exists(regions_file, logger)
dir_exists(os.path.dirname(output), logger)
if args.dataloader is not None:
Dl = kipoi.get_dataloader_factory(args.dataloader,
args.dataloader_source)
else:
Dl = model.default_dataloader
if not isinstance(args.scores, list):
args.scores = [args.scores]
# TODO - why is this function not a method of the model class?
dts = get_scoring_fns(model, args.scores, args.score_kwargs)
# Load effect prediction related model info
model_info = kipoi_veff.ModelInfoExtractor(model, Dl)
manual_seq_len = args.seq_length
# Select the appropriate region generator and vcf or bed file input
args.file_format = regions_file.split(".")[-1]
bed_region_file = None
vcf_region_file = None
bed_to_region = None
vcf_to_region = None
if args.file_format == "vcf" or regions_file.endswith("vcf.gz"):
vcf_region_file = regions_file
if model_info.requires_region_definition:
# Select the SNV-centered region generator
vcf_to_region = kipoi_veff.SnvCenteredRg(model_info,
seq_length=manual_seq_len)
logger.info('Using variant-centered sequence generation.')
elif args.file_format == "bed":
if model_info.requires_region_definition:
# Select the SNV-centered region generator
bed_to_region = kipoi_veff.BedOverlappingRg(
model_info, seq_length=manual_seq_len)
logger.info('Using bed-file based sequence generation.')
bed_region_file = regions_file
else:
raise Exception("")
if model_info.use_seq_only_rc:
logger.info(
'Model SUPPORTS simple reverse complementation of input DNA sequences.'
)
else:
logger.info(
'Model DOES NOT support simple reverse complementation of input DNA sequences.'
)
from kipoi_veff.mutation_map import _generate_mutation_map
mdmm = _generate_mutation_map(
model,
Dl,
vcf_fpath=vcf_region_file,
bed_fpath=bed_region_file,
batch_size=args.batch_size,
num_workers=args.num_workers,
dataloader_args=dataloader_arguments,
vcf_to_region=vcf_to_region,
bed_to_region=bed_to_region,
evaluation_function_kwargs={'diff_types': dts},
)
mdmm.save_to_file(output)
logger.info('Successfully generated mutation map data')
def score_variants(
model,
dl_args,
input_vcf,
output_vcf=None,
output_writers=None,
scores=["logit_ref", "logit_alt", "ref", "alt", "logit", "diff"],
score_kwargs=None,
num_workers=0,
batch_size=32,
source='kipoi',
seq_length=None,
std_var_id=False,
restriction_bed=None,
return_predictions=False,
model_outputs=None):
"""Score variants: annotate the vcf file using model predictions for the reference and alternative alleles
The functional elements that generate a score from a set of predictions for reference and
alternative allele are defined in the `scores` argument.
This function is the python version of the command-line call `score_variants` and is a convenience version
of the `predict_snvs` function:
Prediction of effects of SNV based on a VCF. If desired the VCF can be stored with the predicted values as
annotation. For a detailed description of the requirements in the yaml files please take a look at
the core `kipoi` documentation on how to write a `dataloader.yaml` file or at the documentation of
`kipoi-veff` in the section: `overview/#model-and-dataloader-requirements`.
# Arguments
model: model string or a model class instance
dl_args: dataloader arguments as a dictionary
input_vcf: input vcf file path
output_vcf: output vcf file path
output_writers: output writers a list of used output writers
scores: list of score names to compute. See `kipoi_veff.scores`
score_kwargs: optional, list of kwargs that corresponds to the entries in score.
num_workers: number of paralell workers to use for dataloading
batch_size: batch_size for dataloading
source: model source name
std_var_id: If true then variant IDs in the annotated VCF will be replaced with a standardised, unique ID.
seq_length: If model accepts variable input sequence length then this value has to be set!
restriction_bed: If dataloader can be run with regions generated from the VCF then only variants that overlap
regions defined in `restriction_bed` will be tested.
return_predictions: return generated predictions also as pandas dataframe.
model_outputs: If set then either a boolean filter or a named filter for model outputs that are reported.
# Returns
dict: containing a pandas DataFrame containing the calculated values
for each model output (target) column VCF SNV line. If `return_predictions == False`, returns None.
"""
import kipoi
in_vcf_path_abs = os.path.realpath(input_vcf)
if isinstance(model, str):
model = kipoi.get_model(model, source=source, with_dataloader=True)
Dataloader = model.default_dataloader
vcf_path_tbx = ensure_tabixed_vcf(
in_vcf_path_abs) # TODO - run this within the function
if output_writers is None:
output_writers = []
if output_vcf is not None:
out_vcf_path_abs = os.path.realpath(output_vcf)
output_writers.append(
VcfWriter(model,
in_vcf_path_abs,
out_vcf_path_abs,
standardise_var_id=std_var_id))
else:
if not output_writers:
raise ValueError(
"Either output_vcf or output_writers need to be specified")
dts = get_scoring_fns(model, scores, score_kwargs)
# Load effect prediction related model info
model_info = kipoi_veff.ModelInfoExtractor(model, Dataloader)
vcf_to_region = _get_vcf_to_region(model_info, restriction_bed, seq_length)
return predict_snvs(model,
Dataloader,
vcf_path_tbx,
batch_size=batch_size,
dataloader_args=dl_args,
num_workers=num_workers,
vcf_to_region=vcf_to_region,
evaluation_function_kwargs={
'diff_types': dts,
'output_filter': model_outputs
},
sync_pred_writer=output_writers,
return_predictions=return_predictions)
