def test_sandbag_min():
print("")
print("")
print("## Testing correctness of sandbag()")
print("")
print("# Testing algorithm on minimal data unjitted")
print("")
settings.enable_jit = False
params = {
'data': min_ref_mat.T,
'annotation': min_ref_cats,
'gene_names': min_ref_gene_names,
'sample_names': min_ref_sample_names
}
marker_pairs = pairs.sandbag(**params)
assert utils.same_marker(marker_pairs, min_ref)
utils.benchmark_test(pairs.sandbag, params)
print("")
print("# Testing algorithm on minimal data jitted, single core")
print("")
settings.verbosity = 4
settings.n_jobs = 1
settings.enable_jit = True
marker_pairs = pairs.sandbag(**params)
assert utils.same_marker(marker_pairs, min_ref)
utils.benchmark_test(pairs.sandbag, params)
print("")
print("# Testing algorithm on minimal data jitted, multi core")
print("")
settings.verbosity = 4
settings.n_jobs = 4
settings.enable_jit = True
marker_pairs = pairs.sandbag(**params)
assert utils.same_marker(marker_pairs, min_ref)
utils.benchmark_test(pairs.sandbag, params)
def sandbag(
adata,
annotation,
gene_names,
sample_names,
fraction=0.65,
filter_genes=None,
filter_samples=None):
"""Generate pairs of genes [Scialdone15]_ [Fechtner18]_.
Calculates the pairs of genes serving as marker pairs for each phase,
based on a matrix of gene counts and an annotation of known phases.
This reproduces the approach of [Scialdone15]_ in the implementation of
[Fechtner18]_.
More information and bug reports `here
<https://github.com/rfechtner/pypairs>`__.
Parameters
----------
adata : :class:`~anndata.AnnData`
The annotated data matrix.
categories : `dict`
Dictionary of lists, i.e. {phase: [sample, ...]},
containing annotation of samples to their phase
gene_names: `list`
List of genes.
sample_names: `list`
List of samples.
fraction : `float`, optional (default: 0.5)
Fraction to be used as threshold.
filter_genes : `list` or `None`, optional (default: `None`)
Genes for sampling the reference set. Default is all genes.
filter_samples : `list` or `None`, optional (default: `None`)
Cells for sampling the reference set. Default is all samples.
Returns
-------
`dict` of `list` of `tuple`, i.e.
{phase: [(Gene1, Gene2), ...]},
containing marker pairs per phase
"""
try:
from pypairs import __version__ as pypairsversion
from distutils.version import LooseVersion
if LooseVersion(pypairsversion) < LooseVersion("v3.0.9"):
raise ImportError('Please only use `pypairs` >= v3.0.9 ')
except ImportError:
raise ImportError('You need to install the package `pypairs`.')
from pypairs.pairs import sandbag
from . import settings
from pypairs import settings as pp_settings
pp_settings.verbosity = settings.verbosity
pp_settings.n_jobs = settings.n_jobs
pp_settings.writedir = settings.writedir
pp_settings.cachedir = settings.cachedir
pp_settings.logfile = settings.logfile
return sandbag(
data = adata,
annotation = annotation,
gene_names = gene_names,
sample_names = sample_names,
fraction = fraction,
filter_genes = filter_genes,
filter_samples = filter_samples
)
def sandbag(
adata: Union[AnnData],
annotation: Optional[Mapping[str, Genes]] = None,
*,
fraction: float = 0.65,
filter_genes: Optional[Genes] = None,
filter_samples: Optional[Genes] = None,
) -> Dict[str, List[Tuple[str, str]]]:
"""\
Calculate marker pairs of genes. [Scialdone15]_ [Fechtner18]_.
Calculates the pairs of genes serving as marker pairs for each phase,
based on a matrix of gene counts and an annotation of known phases.
This reproduces the approach of [Scialdone15]_ in the implementation of
[Fechtner18]_.
More information and bug reports `here
<https://github.com/rfechtner/pypairs>`__.
Parameters
----------
adata
The annotated data matrix.
annotation
Mapping from category to genes, e.g. `{'phase': [Gene1, ...]}`.
Defaults to ``data.vars['category']``.
fraction
Fraction of cells per category where marker criteria must be satisfied.
filter_genes
Genes for sampling the reference set. Defaults to all genes.
filter_samples
Cells for sampling the reference set. Defaults to all samples.
Returns
-------
A dict mapping from category to lists of marker pairs, e.g.:
`{'Category_1': [(Gene_1, Gene_2), ...], ...}`.
Examples
--------
>>> from scanpy.external.tl import sandbag
>>> from pypairs import datasets
>>> adata = datasets.leng15()
>>> marker_pairs = sandbag(adata, fraction=0.5)
"""
_check_import()
from pypairs.pairs import sandbag
from pypairs import settings as pp_settings
pp_settings.verbosity = settings.verbosity
pp_settings.n_jobs = settings.n_jobs
pp_settings.writedir = settings.writedir
pp_settings.cachedir = settings.cachedir
pp_settings.logfile = settings.logfile
return sandbag(
data=adata,
annotation=annotation,
fraction=fraction,
filter_genes=filter_genes,
filter_samples=filter_samples,
)
def test_sandbag_inputs():
print("")
print("")
print("## Testing different input types for sandbag()")
settings.n_jobs = 4
settings.verbosity = 4
print("")
print("# Testing AnnData obj, including annotation")
print("")
training_data = datasets.leng15(mode='sorted', gene_sub=list(range(0, 1000)))
marker_pairs = pairs.sandbag(training_data)
assert utils.same_marker(marker_pairs, ref_markers)
print("")
print("# Testing AnnData obj, with separate annotation")
print("")
annotation = {
cat: [i for i, x in enumerate(training_data.obs['category']) if x == cat]
for cat in ["G1", "S", "G2M"]
}
marker_pairs = pairs.sandbag(training_data, annotation=annotation)
assert utils.same_marker(marker_pairs, ref_markers)
print("")
print("# Testing DataFrame obj, with separate annotation")
print("")
training_data_df = DataFrame(training_data.X)
sample_names = list(training_data.obs_names)
gene_names = list(training_data.var_names)
training_data_df.Index = sample_names
training_data_df.columns = gene_names
marker_pairs = pairs.sandbag(training_data_df, annotation=annotation)
assert utils.same_marker(marker_pairs, ref_markers)
print("")
print("# Testing DataFrame obj, with separate annotation and separate gene-/sample_names")
print("")
marker_pairs = pairs.sandbag(training_data_df, annotation, gene_names, sample_names)
assert utils.same_marker(marker_pairs, ref_markers)
print("")
print("# Testing ndarray obj, with separate annotation and separate gene-/sample_names")
print("")
training_data_np = training_data_df.values
marker_pairs = pairs.sandbag(training_data_np, annotation, gene_names, sample_names)
assert utils.same_marker(marker_pairs, ref_markers)
