import gzip
from io import StringIO
import itertools
from pkg_resources import resource_filename
from nibabel.volumeutils import Recoder
import numpy as np
import pandas as pd
from abagen import io, utils
from abagen.datasets import _fetch_alleninf_coords
# AHBA structure IDs corresponding to different brain parts
ONTOLOGY = Recoder(
(('4008', 'cerebral cortex', 'cortex'),
('4275', 'cerebral nuclei', 'subcortex'),
('4391', 'diencephalon', 'subcortex'),
('9001', 'mesencephalon', 'subcortex'),
('4696', 'cerebellum', 'cerebellum'),
('4833', 'metencephalon', 'cerebellum'),
('9512', 'myelencephalon', 'cerebellum')),
fields=('id', 'name', 'structure')
)
def reannotate_probes(probes):
"""
Replaces gene symbols in `probes` with reannotated data
Uses annotations from [1]_ to replace probe annotations shipped with AHBA
data. Any probes that were unable to be matched to a gene in reannotation
procedure are not retained.
Parameters
"""
import os
from pkg_resources import resource_filename
from nibabel.volumeutils import Recoder
from nilearn.datasets.utils import _fetch_files
import pandas as pd
from sklearn.utils import Bunch
from abagen import io
WELL_KNOWN_IDS = Recoder((('9861', 'H0351.2001', '178238387', '157722636'),
('10021', 'H0351.2002', '178238373', '157723301'),
('12876', 'H0351.1009', '178238359', '157722290'),
('15496', 'H0351.1015', '178238266', '162021642'),
('14380', 'H0351.1012', '178238316', '157721937'),
('15697', 'H0351.1016', '178236545', '157682966')),
fields=(
'subj',
'uid',
'url',
't1w',
))
VALID_DONORS = sorted(
WELL_KNOWN_IDS.value_set('subj') | WELL_KNOWN_IDS.value_set('uid'))
def _get_dataset_dir(dataset_name, data_dir=None, verbose=1):
"""
Gets path to `dataset_name`
Parameters
# caveat: Note that it's ambiguous to get the code given the bytespervoxel
# caveat 2: Note that the bytespervox you get is in str ( not an int)
_dtdefs = ( # code, conversion function, dtype, bytes per voxel
(0, 'uint8', '>u1', '1', 'MRI_UCHAR', np.uint8, np.dtype(np.uint8),
np.dtype(np.uint8).newbyteorder('>')),
(4, 'int16', '>i2', '2', 'MRI_SHORT', np.int16, np.dtype(np.int16),
np.dtype(np.int16).newbyteorder('>')),
(1, 'int32', '>i4', '4', 'MRI_INT', np.int32, np.dtype(np.int32),
np.dtype(np.int32).newbyteorder('>')),
(3, 'float', '>f4', '4', 'MRI_FLOAT', np.float32, np.dtype(np.float32),
np.dtype(np.float32).newbyteorder('>')))
# make full code alias bank, including dtype column
data_type_codes = Recoder(_dtdefs,
fields=('code', 'label', 'dtype', 'bytespervox',
'mritype', 'np_dtype1', 'np_dtype2',
'numpy_dtype'))
class MGHError(Exception):
"""Exception for MGH format related problems.
To be raised whenever MGH is not happy, or we are not happy with
MGH.
"""
pass
class MGHHeader(object):
'''
The header also consists of the footer data which MGH places after the data
# vi: set ft=python sts=4 ts=4 sw=4 et:
### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ##
#
# See COPYING file distributed along with the NiBabel package for the
# copyright and license terms.
#
### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ##
from nibabel.volumeutils import Recoder
import numpy as np
xform_codes = Recoder(
( # code, label
(0, 'unknown',
"NIFTI_XFORM_UNKNOWN"), # Code for transform unknown or absent
(1, 'scanner', "NIFTI_XFORM_SCANNER_ANAT"),
(2, 'aligned', "NIFTI_XFORM_ALIGNED_ANAT"),
(3, 'talairach', "NIFTI_XFORM_TALAIRACH"),
(4, 'mni', "NIFTI_XFORM_MNI_152")),
fields=('code', 'label', 'niistring'))
try:
_float128t = np.float128
except AttributeError:
_float128t = np.void
try:
_complex256t = np.complex256
except AttributeError:
_complex256t = np.void
dtdefs = ( # code, label, dtype definition, niistring, XXX: format for store in txt
def _get_unionization_from_experiment(experiment_id,
structures=None,
attributes=None,
average=True,
verbose=False):
"""
Gets unionization data for provided experiment(s) `experiment_id`
Parameters
----------
experiment_id : int or list
Numerical experiment ID. If multiple experiments are provided the
requested `attributes` will be averaged across experiments
structures : list, optional
List of structures (id, acronym, or name) for which to get unionization
information associated with provided `experiment_id`. If not specified
uses structures documented in [1]_. Specifying either the id or name is
recommended as acronyms are not unique to structures. Default: None
attributes : str or list, optional
Which attributes / information to obtain for the provided structure.
See :func:`abagen.mouse.available_unionization_info` for list of
available attributes to request. If not specified all available
attributes will be returned. Default: None
average : bool, optional
Whether to average across experiments if `experiment_id` is provided as
a list. Only experiments probing the same gene will be considered for
averaging. Default: True
verbose : bool, optional
Whether to print status messages. Default: False
Returns
-------
unionization : pandas.DataFrame
Where columns are unionization attributes and the index corresponds to
gene ids and strucuture ids
References
----------
.. [1] Rubinov, M., Ypma, R. J., Watson, C., & Bullmore, E. T. (2015).
Wiring cost and topological participation of the mouse brain connectome.
Proceedings of the National Academy of Sciences, 112(32), 10032-10037.
"""
if isinstance(experiment_id, (str, int)):
experiment_id = [experiment_id]
if structures is None:
# read default structure list (from Rubinov et al., 2015, PNAS)
structures = fetch_rubinov2015_structures(entry_type='id')
elif isinstance(structures, (str, int)):
structures = [structures]
# we need to coerce all provided structures to be integer ids, NOT strings
# so fetch all available structures then recode them to ids
if any(isinstance(f, str) for f in structures):
structs = np.asarray(fetch_allenref_structures(verbose=False))
structs = Recoder(structs.tolist(), fields=['acronym', 'id', 'name'])
structures = list(set(structs.id.get(f) for f in structures))
# determine which attributes to request; if we don't have to request all
# of them then we can speed up the API call
if attributes is None:
attributes = ['expression_density']
elif attributes == 'all':
attributes = _UNIONIZATION_ATTRIBUTES
elif isinstance(attributes, str):
attributes = [attributes]
includes = ['structure_unionizes', 'genes']
criteria = [
'[id$in{}]'.format(','.join([str(f) for f in experiment_id])),
'products[id$eq1]',
'structure_unionizes[structure_id$in{}]'.format(','.join(
[str(f) for f in structures])),
]
req_attributes = [
'id', 'structure_unionizes', 'structure_unionizes.structure_id'
] + ['structure_unionizes.' + attr for attr in attributes]
info = _make_api_query('SectionDataSet',
includes=includes,
criteria=criteria,
attributes=req_attributes,
verbose=verbose)
for n, exp in enumerate(info):
keep = exp['structure_unionizes']
for struc in keep:
struc['gene_id'] = exp['genes'][0]['id']
struc['experiment_id'] = exp['id']
info[n] = keep
# construct data frame from requested unionization info
info = pd.DataFrame(list(itertools.chain.from_iterable(info)))
if average:
info = info.groupby(['gene_id', 'structure_id']).mean()
else:
info = info.set_index(['gene_id', 'experiment_id', 'structure_id'])
return info[attributes]
def get_unionization_from_gene(id=None,
acronym=None,
name=None,
slicing_direction='sagittal',
structures=None,
attributes=None,
average=True,
verbose=False):
"""
Gets unionization data for provided gene(s)
One of `id`, `acronym`, or `name` must be provided.
Parameters
----------
id : int, optional
Numerical gene ID
acronym : str, optional
Short-form gene acronym (case sensitive)
name : str, optional
Full gene name (case sensitive)
slicing_direction : {'sagittal', 'coronal'}, optional
Slicing direction of brain tissue
structures : list, optional
List of structures (id, acronym, or name) for which to get unionization
information associated with provided `experiment_id`. If not specified
uses structures documented in [1]_. Specifying either the id or name is
recommended as acronyms are not unique to structures. Default: None
attributes : str or list, optional
Which attributes / information to obtain for the provided gene. See
:func:`abagen.mouse.available_gene_info` for list of available
attributes to request. If not specified then only 'expression_density'
will be returned. Specifying 'all' will return all information.
Default: None
average : bool, optional
Whether to average across experiments if there are multiple experiments
corresponding to any provided gene(s). Only experiments probing the
same gene will be considered for averaging, and distinct structures
will be retained. Default: True
verbose : bool, optional
Whether to print status messages. Default: False
Returns
-------
unionization : pandas.DataFrame
Where columns are unionization attributes and the index corresponds to
strucuture and gene ids (if `experiments` is provided as a list
with multiple genes). If `average=False`, `experiments` will also be a
level in index
Examples
--------
>>> from abagen import mouse
>>> mouse.get_unionization_from_gene(acronym='Pdyn', structures=[22, 31])
expression_density
gene_id structure_id
18376 22 0.024840
31 0.017199
>>> mouse.get_unionization_from_gene(acronym=['Ace', 'Cd99'], structures=[22, 31])
expression_density
gene_id structure_id
11210 22 0.001283
31 0.001427
163028 22 0.067537
31 0.056442
References
----------
.. [1] Rubinov, M., Ypma, R. J., Watson, C., & Bullmore, E. T. (2015).
Wiring cost and topological participation of the mouse brain connectome.
Proceedings of the National Academy of Sciences, 112(32), 10032-10037.
""" # noqa
directions = ['sagittal', 'coronal']
if slicing_direction not in directions:
raise ValueError(
'Slicing_direction {} is invalid. Must be in {}.'.format(
slicing_direction, directions))
if structures is None:
# read default structure list (from Rubinov et al., 2015, PNAS)
structures = fetch_rubinov2015_structures(entry_type='id')
elif isinstance(structures, (str, int)):
structures = [structures]
# we need to coerce all provided structures to be integer ids, NOT strings
# so fetch all available structures then recode them to ids
if any(isinstance(f, str) for f in structures):
structs = np.asarray(fetch_allenref_structures(verbose=False))
structs = Recoder(structs.tolist(), fields=['acronym', 'id', 'name'])
structures = list(set(structs.id.get(f) for f in structures))
exp_ids = _get_experiments_from_gene(id=id,
acronym=acronym,
name=name,
slicing_direction=slicing_direction,
verbose=verbose)
data = _get_unionization_from_experiment(exp_ids,
structures=structures,
attributes=attributes,
average=average,
verbose=verbose)
return data
from nibabel.volumeutils import Recoder
import numpy as np
import pandas as pd
from scipy.spatial.distance import cdist
from . import io, utils
# AHBA structure IDs corresponding to different brain parts
ONTOLOGY = Recoder(
(('4008', 'cerebral cortex', 'cortex'),
('4275', 'cerebral nuclei', 'subcortex'),
('4391', 'diencephalon', 'subcortex'),
('9001', 'mesencephalon', 'subcortex'),
('4696', 'cerebellum', 'cerebellum'),
('9131', 'pons', 'brainstem'),
('9512', 'myelencephalon', 'brainstem'),
('9218', 'white matter', 'white matter'),
('9352', 'sulci & spaces', 'other'),
('4219', 'hippocampal formation', 'subcortex')),
fields=('id', 'name', 'structure')
)
def update_mni_coords(annotation):
"""
Replaces MNI coords in `annotation` with corrected coords from `alleninf`
Parameters
----------
annotation : str
"""Read/write linear transforms."""
import numpy as np
from nibabel.volumeutils import Recoder
from nibabel.affines import voxel_sizes
from .base import BaseLinearTransformList, StringBasedStruct, TransformFileError
transform_codes = Recoder(
(
(0, "LINEAR_VOX_TO_VOX"),
(1, "LINEAR_RAS_TO_RAS"),
(2, "LINEAR_PHYSVOX_TO_PHYSVOX"),
(14, "REGISTER_DAT"),
(21, "LINEAR_COR_TO_COR"),
),
fields=("code", "label"),
)
class VolumeGeometry(StringBasedStruct):
"""Data structure for regularly gridded images."""
template_dtype = np.dtype(
[
("valid", "i4"), # Valid values: 0, 1
("volume", "i4", (3, 1)), # width, height, depth
("voxelsize", "f4", (3, 1)), # xsize, ysize, zsize
("xras", "f8", (3, 1)), # x_r, x_a, x_s
("yras", "f8", (3, 1)), # y_r, y_a, y_s
("zras", "f8", (3, 1)), # z_r, z_a, z_s
"""Read/write linear transforms."""
import numpy as np
from nibabel.volumeutils import Recoder
from nibabel.affines import voxel_sizes
from .base import BaseLinearTransformList, StringBasedStruct, TransformFileError
transform_codes = Recoder((
(0, 'LINEAR_VOX_TO_VOX'),
(1, 'LINEAR_RAS_TO_RAS'),
(2, 'LINEAR_PHYSVOX_TO_PHYSVOX'),
(14, 'REGISTER_DAT'),
(21, 'LINEAR_COR_TO_COR')),
fields=('code', 'label'))
class VolumeGeometry(StringBasedStruct):
"""Data structure for regularly gridded images."""
template_dtype = np.dtype([
('valid', 'i4'), # Valid values: 0, 1
('volume', 'i4', (3, 1)), # width, height, depth
('voxelsize', 'f4', (3, 1)), # xsize, ysize, zsize
('xras', 'f8', (3, 1)), # x_r, x_a, x_s
('yras', 'f8', (3, 1)), # y_r, y_a, y_s
('zras', 'f8', (3, 1)), # z_r, z_a, z_s
('cras', 'f8', (3, 1)), # c_r, c_a, c_s
('filename', 'U1024')]) # Not conformant (may be >1024 bytes)
dtype = template_dtype