def list_filters(mart_name, dataset):
"""
List filters for mart name and dataset
>>> mart_name = 'ensembl'
>>> dataset = 'dmelanogaster_gene_ensembl'
>>> list_filters(mart_name, dataset)[:3]
name description
0 chromosome_name Chromosome name
1 start Gene Start (bp)
2 end Gene End (bp)
<BLANKLINE>
[3 rows x 2 columns]
"""
dataset = r.useDataset(dataset, mart=r.useMart(mart_name))
return rpy2_to_pandas(r.listFilters(dataset))
def list_attributes(mart_name, dataset):
"""
Returns a pandas.DataFrame listing attributes for mart name and dataset
>>> mart_name = 'ensembl'
>>> dataset = 'dmelanogaster_gene_ensembl'
>>> list_attributes(mart_name, dataset)[:3]
name description
0 ensembl_gene_id Ensembl Gene ID
1 ensembl_transcript_id Ensembl Transcript ID
2 ensembl_peptide_id Ensembl Protein ID
<BLANKLINE>
[3 rows x 2 columns]
"""
dataset = r.useDataset(dataset, mart=r.useMart(mart_name))
return rpy2_to_pandas(r.listAttributes(dataset))
def make_lookup(mart_name, dataset, attributes, filters=None, values=None,
unique_rows=True):
"""
Given a mart name, dataset name, and a list of attributes, return
a pandas.DataFrame indexed by the first attribute in the list provided.
In R, filters is a character vector, and values is either a single
character vector (if only one filter provided) or a list of character
vectors.
This function allows `filters` to be a dictionary where keys are filters
and values are...values.
>>> mart_name = 'ensembl'
>>> dataset = 'dmelanogaster_gene_ensembl'
>>> filters = ['flybase_gene_id', 'chromosome_name']
>>> attributes = ['flybase_gene_id', 'flybasename_gene', 'chromosome_name']
>>> values = [['FBgn0031208', 'FBgn0002121', 'FBgn0031209', 'FBgn0051973'], ['2L']]
>>> df = make_lookup(
... mart_name=mart_name,
... dataset=dataset,
... attributes=attributes,
... filters=filters,
... values=values)
Alternatively, make a dictionary of filters: values, in which case you
don't need to provide `values` separately:
>>> filters = {
... 'flybase_gene_id': ['FBgn0031208', 'FBgn0002121', 'FBgn0031209', 'FBgn0051973'],
... 'chromosome_name': ['2L']}
>>> df2 = make_lookup(
... mart_name=mart_name,
... dataset=dataset,
... attributes=attributes,
... filters=filters)
Confirm that both methods yield identical results:
>>> assert np.all(df.values == df2.values)
Check results:
>>> df.head()
flybasename_gene chromosome_name
flybase_gene_id
FBgn0002121 l(2)gl 2L
FBgn0031208 CG11023 2L
FBgn0031209 Ir21a 2L
FBgn0051973 Cda5 2L
<BLANKLINE>
[4 rows x 2 columns]
Indexing by gene ID (or whatever was the first attribute provided):
>>> df.ix['FBgn0031209']
flybasename_gene Ir21a
chromosome_name 2L
Name: FBgn0031209, dtype: object
Extracting data:
>>> df.ix['FBgn0031209']['flybasename_gene']
'Ir21a'
Or get all names:
>>> df['flybasename_gene']
flybase_gene_id
FBgn0002121 l(2)gl
FBgn0031208 CG11023
FBgn0031209 Ir21a
FBgn0051973 Cda5
Name: flybasename_gene, dtype: object
"""
mart = r.useDataset(dataset, mart=r.useMart(mart_name))
attributes = robjects.StrVector(attributes)
kwargs = dict(
attributes=attributes,
uniqueRows=unique_rows,
mart=mart
)
def _filter_and_values_to_RList(d):
"""`d` is a dictionary of filters: values. Returns a StrVector and
a ListVector of StrVectors"""
# Could use ListVector directly with the dict, but want to guarantee
# positional order of filters and values
f = robjects.StrVector(list(d.keys()))
v = robjects.ListVector(
rpy2.rlike.container.TaggedList(
d.values(),
tags=list(d.keys())
)
)
return f, v
if isinstance(filters, dict):
if values is not None:
raise ValueError("`values` are already specified in the "
"`filters` dictionary")
filter_value_dict = filters
_filters, _values = _filter_and_values_to_RList(filter_value_dict)
kwargs['filters'] = _filters
kwargs['values'] = _values
elif filters is None:
if values is not None:
raise ValueError("`filters` must be specified if `values` "
"is specified; alternatively use a dictionary "
" for `filters`")
elif filters and values:
# values needs to be a list of lists; convert it to one if it's not
# already
if not isinstance(values[0], (list, tuple)):
values = [values]
# If we got here, then assume filters is a list or tuple
if len(filters) != len(values):
raise ValueError('Length of `filters` and `values` must match')
filter_value_dict = dict(zip(filters, values))
_filters, _values = _filter_and_values_to_RList(filter_value_dict)
kwargs['filters'] = _filters
kwargs['values'] = _values
else:
raise ValueError('unhandled case')
results = r.getBM(**kwargs)
return rpy2_to_pandas(results, index_col=0)