def sandbag(adata,
phases,
fraction=0.5,
subset_genes=None,
subset_samples=None,
processes=1):
"""Generate pairs of genes [Scialdone15]_.
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 in Pairs [Scialdone15]_ and has been
implemented for Scanpy by Ron Fechtner.
Parameters
----------
adata : :class:`~scanpy.api.AnnData`
The annotated data matrix.
phases : `dict`
Dictionary of lists, i.e. {phase: [sample, ...]},
containing annotation of samples to their phase
fraction : `float`, optional (default: 0.5)
Fraction to be used as threshold.
subset_genes : `list` or `None`, optional (default: `None`)
Genes for sampling the reference set. Default is all genes.
subset_samples : `list` or `None`, optional (default: `None`)
Cells for sampling the reference set. Default is all samples.
processes : `int`, optional (default: 1)
Number of concurrent processes to be used. 0 = use all available cores.
Returns
-------
`dict` of `list` of `tuple`, i.e.
{phase: [(Gene1, Gene2), ...]},
containing marker pairs per phase
Examples
--------
See this `notebook <https://github.com/theislab/scanpy_usage/tree/master/180209_cell_cycle>`_.
"""
try:
import pypairs
except ImportError:
raise ImportError('You need to install the package `pypairs`.')
x = pd.DataFrame(adata.X)
return pypairs.sandbag(x=x,
phases=phases,
subset_genes=subset_genes,
subset_samples=subset_samples,
processes=processes)
def sandbag(adata,
phases,
fraction=0.5,
subset_genes=None,
subset_samples=None,
n_jobs=1):
"""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.
phases : `dict`
Dictionary of lists, i.e. {phase: [sample, ...]},
containing annotation of samples to their phase
fraction : `float`, optional (default: 0.5)
Fraction to be used as threshold.
subset_genes : `list` or `None`, optional (default: `None`)
Genes for sampling the reference set. Default is all genes.
subset_samples : `list` or `None`, optional (default: `None`)
Cells for sampling the reference set. Default is all samples.
n_jobs : `int`, optional (default: 1)
Number of concurrent n_jobs to be used. 0 = use all available cores.
Returns
-------
`dict` of `list` of `tuple`, i.e.
{phase: [(Gene1, Gene2), ...]},
containing marker pairs per phase
"""
try:
import pypairs
except ImportError:
raise ImportError('You need to install the package `pypairs`.')
x = pd.DataFrame(adata.X)
return pypairs.sandbag(x=x,
phases=phases,
subset_genes=subset_genes,
subset_samples=subset_samples,
processes=n_jobs)
def load_ocope_marker(relPath,
fraction=0.6,
cc_only=True,
weighted=False,
triplets=False):
# Load matrix
gencounts_oscope = pandas.read_csv(
Path(relPath + "GSE64016_H1andFUCCI_normalized_EC_human.csv"))
# Set index right
gencounts_oscope.set_index("Unnamed: 0", inplace=True)
# Subset sorted
gencounts_oscope_sorted = gencounts_oscope.iloc[:, [
gencounts_oscope.columns.get_loc(c) for c in gencounts_oscope.columns
if "G1_" in c or "G2_" in c or "S_" in c
]]
# Define annotation
is_G1 = [
gencounts_oscope_sorted.columns.get_loc(c)
for c in gencounts_oscope_sorted.columns if "G1_" in c
]
is_S = [
gencounts_oscope_sorted.columns.get_loc(c)
for c in gencounts_oscope_sorted.columns if "S_" in c
]
is_G2M = [
gencounts_oscope_sorted.columns.get_loc(c)
for c in gencounts_oscope_sorted.columns if "G2_" in c
]
annotation = {"G1": list(is_G1), "S": list(is_S), "G2M": list(is_G2M)}
go_0007049 = [
line.replace("\n", "").replace("\r", "")
for line in open(relPath + "go_0007049_homoSapiens.csv", "r")
]
cycle_base = [
line.split("\t")[0] for i, line in enumerate(
open(relPath + "cyclebase_top1000_genes.tsv", "r")) if 0 < i
]
cycle_genes = numpy.unique(numpy.concatenate((go_0007049, cycle_base), 0))
if cc_only:
return pairs.sandbag(gencounts_oscope_sorted,
phases=annotation,
subset_genes=list(cycle_genes),
fraction=fraction,
processes=10,
verbose=True,
weighted=weighted,
triplets=triplets)
else:
return pairs.sandbag(gencounts_oscope_sorted,
phases=annotation,
fraction=fraction,
processes=10,
verbose=True,
weighted=weighted,
triplets=triplets)