def get_datasets(mart):
bm = BioMart(verbose=False)
datasets = bm.datasets(mart, raw=True)
return pd.read_table(StringIO(datasets), header=None, usecols=[1, 2],
names = ["Name", "Description"])
def get_attributes(dataset):
bm = BioMart(verbose=False)
attributes = bm.attributes(dataset)
attr_dicts = [{"Attribute": k, "Description": v[0]}
for k, v in attributes.items()]
return pd.DataFrame.from_dict(attr_dicts)
def query_biomart(self,
dataset,
attributes,
host="www.ensembl.org",
cache=True,
save_filename=None):
bm = BioMart(host=host)
bm.new_query()
bm.add_dataset_to_xml(dataset)
for at in attributes:
bm.add_attribute_to_xml(at)
xml_query = bm.get_xml()
print("Querying {} from {} with attributes {}...".format(
dataset, host, attributes))
results = bm.query(xml_query)
df = pd.read_csv(StringIO(results),
header=None,
names=attributes,
sep="\t",
low_memory=True)
if cache:
self.cache_dataset(dataset, df, save_filename)
return df
def test_biomart_constructor():
s = BioMart()
try:
s.registry()
except:
pass
try:
s.host = "dummy"
except:
pass
s.host = "www.ensembl.org"
def test_biomart_constructor():
s = BioMart()
try:
s.registry()
except:
pass
try:
s.host = "dummy"
except:
pass
s.host = "www.ensembl.org"
def run(self, inputs, outputs):
"""Run the analysis."""
for exp in inputs.expressions:
if exp.output.species != inputs.expressions[0].output.species:
self.error(
"Input samples are of different Species: "
f"{exp.output.species} and {inputs.expressions[0].output.species}."
)
if exp.output.exp_type != inputs.expressions[0].output.exp_type:
self.error(
"Input samples have different Normalization types: "
f"{exp.output.exp_type} and {inputs.expressions[0].output.exp_type}."
)
if exp.output.platform != inputs.expressions[0].output.platform:
self.error(
"Input samples have different Microarray platform types: "
f"{exp.output.platform} and {inputs.expressions[0].output.platform}."
)
if exp.output.platform_id != inputs.expressions[
0].output.platform_id:
self.error(
"Input samples have different GEO platform IDs: "
f"{exp.output.platform_id} and {inputs.expressions[0].output.platform_id}."
)
species = inputs.expressions[0].output.species
platform = inputs.expressions[0].output.platform
platform_id = inputs.expressions[0].output.platform_id
joined_expressions = join_expressions(inputs.expressions)
probe_ids = joined_expressions.index.unique()
if inputs.mapping_file:
mapping_file = inputs.mapping_file.import_file(
imported_format="compressed")
stem = Path(mapping_file).stem
supported_extensions = (".tab", ".tsv", ".txt")
if not stem.endswith(supported_extensions):
self.error(
"Mapping file has unsupported file name extension. "
f"The supported extensions are {supported_extensions}.")
mapping = pd.read_csv(
mapping_file,
sep="\t",
header=0,
names=["ensembl_id", "probe"],
dtype=str,
)
mapping = mapping.drop_duplicates()
if inputs.source:
source = inputs.source
else:
self.error(
"Custom probe id mapping file was provided but no source was selected."
)
if inputs.build:
build = inputs.build
else:
self.error(
"Custom probe id mapping file was provided but genome build was not defined."
)
probe_mapping = "Custom"
else:
if not platform_id:
self.error(
"Custom mapping file should be provided when samples do not have a GEO platform defined"
)
if platform_id not in PLATFORM_MAP:
self.error(f"GEO platform {platform_id} is not supported.")
species_low = species.lower()
dataset = f"{species_low[0]}{species_low.split(' ')[1]}_gene_ensembl"
probe_mapping = PLATFORM_MAP[platform_id]
try:
b = BioMart()
except IOError:
raise Exception(
"None of the ENSEMBL Biomart hosts is reachable.")
except Exception as e:
raise Exception(f"Unexpected biomart error: {e}")
b.add_dataset_to_xml(dataset)
b.add_attribute_to_xml("ensembl_gene_id")
b.add_attribute_to_xml(probe_mapping) # type of microarray
b.add_filter_to_xml(probe_mapping, ",".join(probe_ids))
xml_query = b.get_xml()
res = b.query(xml_query)
mapping = pd.read_csv(
StringIO(res),
sep="\t",
header=None,
names=["ensembl_id", "probe"],
dtype=str,
)
mapping = mapping.drop_duplicates()
mapping_file = f"{platform}_mapping.tsv"
mapping.to_csv(mapping_file, sep="\t", index=False)
dataset_names = b.get_datasets("ENSEMBL_MART_ENSEMBL")
display_name = dataset_names.loc[
dataset_names["name"] == dataset]["description"].to_string()
# Typical display name would be Human genes (GRCh38.p13)
build = re.search("\((.+?)\)", display_name).group(1)
source = "ENSEMBL"
mapping = mapping.drop_duplicates(subset=["probe"], keep=False)
data = joined_expressions.loc[mapping["probe"]]
data["ensembl_id"] = mapping["ensembl_id"].tolist()
data = data.reset_index()
# For Ensembl IDs with multiple probe IDs retain the one with highest expression.
data["mean"] = data.loc[:,
data.columns.
difference(["probe", "ensembl_id"])].mean(
axis=1)
idx_max = data.groupby(["ensembl_id"])["mean"].idxmax()
data = data.loc[idx_max].set_index("ensembl_id")
data = data.drop(columns=["probe", "mean"])
data.index.name = "Gene"
mapped_file = "mapped_expressions.tsv.gz"
data.to_csv(mapped_file, sep="\t", index=True, compression="gzip")
for column, exp in zip(data.columns, inputs.expressions):
mapped_column = f"{column}_mapped_exp.tsv.gz"
data.to_csv(
mapped_column,
sep="\t",
index=True,
columns=[column],
header=["Expression"],
index_label="Gene",
compression="gzip",
)
self.run_process(
"mapped-microarray-expression",
{
"exp_unmapped": exp.id,
"exp": mapped_column,
"source": source,
"build": build,
"probe_mapping": probe_mapping,
},
)
outputs.mapped_exp = mapped_file
outputs.mapping = mapping_file
outputs.probe_mapping = probe_mapping
outputs.platform = platform
if platform_id:
outputs.platform_id = platform_id
def _test_reactome_example():
# this is not working anymore...
s = BioMart()
s.lookfor("reactome")
s.datasets("REACTOME")
#['interaction', 'complex', 'reaction', 'pathway']
s.new_query()
s.add_dataset_to_xml("pathway")
s.add_filter_to_xml("species_selection", "H**o sapiens")
s.add_attribute_to_xml("pathway_db_id")
s.add_attribute_to_xml("_displayname")
xmlq = s.get_xml()
res = s.query(xmlq)
def test_general(self):
# test another host
s = BioMart(host="www.ensembl.org")
def id_map_ensembl(to_annotation,version,species,psm_protein_id):
'''
:param to_annotation: target identifier annotation (i.e. uniprot_swissprot)
:param version: Database version
:param species: Full species name
:param psm_protein_id: list of IDs to be converted
:return: BioMart results
'''
# If species is in plantsDB, execute plants adjusted function
if species=="arabidopsis_thaliana":
result=id_map_ensembl_plants(to_annotation,version,species,psm_protein_id)
return result
else:
#create connection
query_string=_id_in_xml_query_(psm_protein_id)
version=_get_ensembl_archive_(version,species)
dataset=_get_ensembl_dataset_(species)
biomart = BioMart(host=version)
#add filters
biomart.add_dataset_to_xml(dataset)
biomart.add_filter_to_xml(to_annotation,query_string)
#add attributs
biomart.add_attribute_to_xml("ensembl_transcript_id")
biomart.add_attribute_to_xml("transcript_length")
biomart.add_attribute_to_xml("uniprot_sptrembl")
attributes=biomart.attributes(dataset)
#execute query
xml_query=biomart.get_xml()
result=biomart.query(xml_query)
result=result.split("\n")
return result
def id_map_ensembl_plants(to_annotation,version,species,psm_protein_id):
#create connection
query_string=_id_in_xml_query_(psm_protein_id)
version=_get_ensembl_archive_(version,species)
dataset=_get_ensembl_dataset_(species)
biomart = BioMart(host=version)
#add filters
biomart.add_dataset_to_xml(dataset)
biomart.add_filter_to_xml(to_annotation+"_accession",query_string)
#add attributs
biomart.add_attribute_to_xml("ensembl_transcript_id")
biomart.add_attribute_to_xml("transcript_start")
biomart.add_attribute_to_xml("uniprot_swissprot_accession")
biomart.add_attribute_to_xml("transcript_end")
#execute query
xml_query=biomart.get_xml()
xml_query=xml_query.replace('virtualSchemaName = "default"','virtualSchemaName = "plants_mart_30"')
#parse results and adjust length
temp_result=biomart.query(xml_query).split("\n")
result=[]
for row in temp_result:
items=row.split("\t")
# print row
if len(items)==4:
length=int(items[3])-int(items[1])+1
result.append(items[0]+"\t"+str(length)+"\t"+items[2])
return result
def id_map_ensembl_plants(to_annotation, version, species, psm_protein_id):
'''
:param to_annotation: to which annotation
:param version: ensembl version
:param species: species
:param psm_protein_id: list of protein IDs
:return: list of protein ID's converted to ENSEMBL
'''
#create connection
query_string = _id_in_xml_query_(psm_protein_id)
version = _get_ensembl_archive_(version, species)
dataset = _get_ensembl_dataset_(species)
biomart = BioMart(host=version)
#add filters
biomart.add_dataset_to_xml(dataset)
biomart.add_filter_to_xml(to_annotation + "_accession", query_string)
#add attributs
biomart.add_attribute_to_xml("ensembl_transcript_id")
biomart.add_attribute_to_xml("transcript_start")
biomart.add_attribute_to_xml("uniprot_swissprot_accession")
biomart.add_attribute_to_xml("transcript_end")
#execute query
xml_query = biomart.get_xml()
xml_query = xml_query.replace('virtualSchemaName = "default"',
'virtualSchemaName = "plants_mart_30"')
#parse results and adjust length
temp_result = biomart.query(xml_query).split("\n")
result = []
for row in temp_result:
items = row.split("\t")
# print row
if len(items) == 4:
length = int(items[3]) - int(items[1]) + 1
result.append(items[0] + "\t" + str(length) + "\t" + items[2])
return result
def retrieve_data_from_biomart(version,species,transcript_id,three_frame_translation):
'''
:param version: Database version
:param species: Full species name
:param transcript_id: list of transcript IDs
:return: BioMart results
'''
#create connection
tr_query=_id_in_xml_query_(transcript_id)
version=_get_ensembl_archive_(version,species)
dataset=_get_ensembl_dataset_(species)
biomart = BioMart(host=version)
#add filters
biomart.add_dataset_to_xml(dataset)
biomart.add_filter_to_xml("ensembl_transcript_id",tr_query)
#add attributes
biomart.add_attribute_to_xml('ensembl_transcript_id')
biomart.add_attribute_to_xml("chromosome_name")
biomart.add_attribute_to_xml("strand")
if three_frame_translation=="Y":
biomart.add_attribute_to_xml("cdna")
else:
biomart.add_attribute_to_xml("coding")
attributes=biomart.attributes(dataset)
#execute query
xml_query=biomart.get_xml()
# create bypass for plants database
if species=="arabidopsis_thaliana":
xml_query=xml_query.replace('virtualSchemaName = "default"','virtualSchemaName = "plants_mart_30"')
result=biomart.query(xml_query)
result=result.split("\n")
return result
def setup_class(klass):
klass.s = BioMart(verbose=False)
def id_map_ensembl(to_annotation, version, species, psm_protein_id):
'''
:param to_annotation: target identifier annotation (i.e. uniprot_swissprot)
:param version: Database version
:param species: Full species name
:param psm_protein_id: list of IDs to be converted
:return: BioMart results
'''
# If species is in plantsDB, execute plants adjusted function
if species == "arabidopsis_thaliana":
result = id_map_ensembl_plants(to_annotation, version, species,
psm_protein_id)
return result
else:
#adjust UniProt xml annotation for BioMart version >87
if int(version) > 87 and "uniprot" in to_annotation:
to_annotation = to_annotation.replace('_', '')
#create connection
query_string = _id_in_xml_query_(psm_protein_id)
version = _get_ensembl_archive_(version, species)
dataset = _get_ensembl_dataset_(species)
biomart = BioMart(host=version)
#add filters
biomart.add_dataset_to_xml(dataset)
biomart.add_filter_to_xml(to_annotation, query_string)
#add attributs
biomart.add_attribute_to_xml("ensembl_transcript_id")
biomart.add_attribute_to_xml("transcript_start")
biomart.add_attribute_to_xml("transcript_end")
biomart.add_attribute_to_xml(to_annotation)
attributes = biomart.attributes(dataset)
#execute query
xml_query = biomart.get_xml()
tmp_result = biomart.query(xml_query)
if len(tmp_result) == 1:
print "ERROR: could not convert ID's trough BioMart, " \
"Please check whether Ensembl version/species were correctly supplied"
tmp_result = tmp_result.split("\n")
result = []
if tmp_result != []:
for i in tmp_result:
i = i.split("\t")
if i[0] != "":
result.append([i[0], (int(i[2]) - int(i[1])), i[3]])
else:
result.append(i)
return result
def setup_class(klass):
# ideally we should not provide a host to be more generic
# but this takes lots of time or is simply down.
klass.s = BioMart(host='www.ensembl.org', verbose=False)
klass.mart_test = 'ENSEMBL_MART_ENSEMBL'
def get_bm(intype, outtype, dataset, mart):
"""Queries biomart for data.
Gets the whole map between INTYPE <-> OUTTYPE and caches it so that disk
based lookups are used afterwards."""
bm = BioMart(verbose=False)
bm.new_query()
bm.add_dataset_to_xml(dataset)
bm.add_attribute_to_xml(intype)
bm.add_attribute_to_xml(outtype)
xml_query = bm.get_xml()
results = bm.query(xml_query)
map_df = pd.read_table(StringIO(results), header=None, names=[intype,
outtype])
outfile = _get_data_output_filename(intype, outtype, dataset, mart,
default_cache_path=default_cache_path)
map_df.to_csv(outfile, sep="\t", index=False)
return map_df
def test_general():
s = BioMart()
#s.registry()
s.datasets("ensembl")
s.version("ensembl")
s.attributes("oanatinus_gene_ensembl")
s.filters("oanatinus_gene_ensembl")
s.configuration("oanatinus_gene_ensembl")
xmlq = """<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE Query>
<Query virtualSchemaName = "default" formatter = "TSV" header = "0" uniqueRows = "0" count = "" datasetConfigVersion = "0.6" >
<Dataset name = "pathway" interface = "default" >
<Filter name = "referencepeptidesequence_uniprot_id_list" value = "P43403"/>
<Attribute name = "stableidentifier_identifier" />
<Attribute name = "pathway_db_id" />
</Dataset>
</Query>
"""
s.query(s._xml_example)
# build own xml using the proper functions
s.add_dataset_to_xml("protein")
s.get_xml()
def biomart():
biomart = BioMart(host='www.ensembl.org', verbose=False)
biomart.mart_test = 'ENSEMBL_MART_ENSEMBL'
return biomart
def biomart():
biomart = BioMart(host='www.ensembl.org', verbose=False)
biomart.mart_test = 'ENSEMBL_MART_ENSEMBL'
return biomart
def _test_reactome_example():
# this is not working anymore...
s = BioMart("reactome.org")
s.lookfor("reactome")
s.datasets("REACTOME")
#['interaction', 'complex', 'reaction', 'pathway']
s.new_query()
s.add_dataset_to_xml("pathway")
s.add_filter_to_xml("species_selection", "H**o sapiens")
s.add_attribute_to_xml("pathway_db_id")
s.add_attribute_to_xml("_displayname")
xmlq = s.get_xml()
res = s.query(xmlq)
def retrieve_data_from_biomart(version, species, transcript_id,
three_frame_translation):
'''
:param version: Database version
:param species: Full species name
:param transcript_id: list of transcript IDs
:return: BioMart results
'''
#create connection
tr_query = _id_in_xml_query_(transcript_id)
version = _get_ensembl_archive_(version, species)
dataset = _get_ensembl_dataset_(species)
biomart = BioMart(host=version)
#add filters
biomart.add_dataset_to_xml(dataset)
biomart.add_filter_to_xml("ensembl_transcript_id", tr_query)
#add attributes
biomart.add_attribute_to_xml('ensembl_transcript_id')
biomart.add_attribute_to_xml("chromosome_name")
biomart.add_attribute_to_xml("strand")
if three_frame_translation == "Y":
biomart.add_attribute_to_xml("cdna")
else:
biomart.add_attribute_to_xml("coding")
attributes = biomart.attributes(dataset)
#execute query
xml_query = biomart.get_xml()
# create bypass for plants database
if species == "arabidopsis_thaliana":
xml_query = xml_query.replace('virtualSchemaName = "default"',
'virtualSchemaName = "plants_mart_30"')
result = biomart.query(xml_query)
result = result.split("\n")
return result
