def cluster(self, strand=None, by=None, **kwargs):
if strand is None:
strand = self.stranded
kwargs = fill_kwargs(kwargs)
kwargs["sparse"] = {"self": False}
_stranded = self.stranded
if not strand and _stranded:
# print(" WOOO " * 100)
self.Strand2 = self.Strand
self = self.unstrand()
if not by:
from pyranges.methods.cluster import _cluster
df = pyrange_apply_single(_cluster, self, strand, kwargs)
else:
from pyranges.methods.cluster import _cluster_by
kwargs["by"] = by
df = pyrange_apply_single(_cluster_by, self, strand, kwargs)
gr = PyRanges(df)
# each cluster got same ids (0 to len). Need to make unique!
new_dfs = {}
first = True
max_id = 0
for k, v in gr.items():
if first:
max_id = v.Cluster.max()
new_dfs[k] = v
first = False
continue
v.loc[:, "Cluster"] += max_id
max_id = v.Cluster.max()
new_dfs[k] = v
if not strand and _stranded:
# print(" wooo " * 100)
# print(new_dfs)
new_dfs = {
k: d.rename(columns={"Strand2": "Strand"})
for k, d in new_dfs.items()
}
# print(new_dfs)
self = PyRanges(new_dfs)
return self
def _to_rle(ranges, value_col=None, strand=True, rpm=False, **kwargs):
try:
from pyrle.methods import coverage
from pyrle import PyRles
except ImportError:
raise Exception(
"Using the coverage method requires that pyrle is installed.")
_kwargs = {
"strand": strand,
"value_col": value_col,
"sparse": {
"self": False
}
} # already sparse
kwargs.update(_kwargs)
result = pyrange_apply_single(coverage, ranges, **kwargs)
if rpm:
multiplier = 1e6 / len(ranges)
result = {k: v * multiplier for k, v in result.items()}
return PyRles(result)
def slack(self, slack):
kwargs = fill_kwargs({"slack": slack})
prg = PyRanges(
pyrange_apply_single(_slack, self, self.stranded, kwargs))
return prg
def _coverage(ranges, value_col=None, strand=True, rpm=False, **kwargs):
try:
from pyrle.methods import coverage
from pyrle import PyRles
except ImportError:
raise Exception(
"Using the coverage method requires that pyrle is installed.")
keep = [value_col if not value_col is None else "Score"]
kwargs = {
"value_col": value_col,
"sparse": {
"self": False
}
} # already sparse
# from pydbg import dbg
result = pyrange_apply_single(coverage, ranges, strand, kwargs)
if rpm:
multiplier = 1e6 / len(ranges)
result = {k: v * multiplier for k, v in result.items()}
return PyRles(result)
def sort(self, by=None, **kwargs):
from pyranges.methods.sort import _sort
kwargs["sparse"] = {"self": False}
if by:
kwargs["by"] = by
kwargs = fill_kwargs(kwargs)
return PyRanges(
pyrange_apply_single(_sort, self, self.stranded, kwargs))
def merge(self, strand=None, **kwargs):
from pyranges.methods.merge import _merge
kwargs["sparse"] = {"self": True}
df = pyrange_apply_single(_merge, self, strand, kwargs)
return PyRanges(df)
def merge(self, strand=None, count=False, **kwargs):
if strand is None:
strand = self.stranded
if not ("by" in kwargs):
kwargs["sparse"] = {"self": True}
from pyranges.methods.merge import _merge
df = pyrange_apply_single(_merge, self, strand, kwargs)
else:
kwargs["sparse"] = {"self": False}
from pyranges.methods.merge import _merge_by
df = pyrange_apply_single(_merge_by, self, strand, kwargs)
if not count:
df = {k: v.drop("Count", axis=1) for k, v in df.items()}
return PyRanges(df)
def tile(self, tile_size, strand=None, **kwargs):
from pyranges.methods.windows import _tiles
kwargs["sparse"] = {"self": False}
kwargs["tile_size"] = tile_size
df = pyrange_apply_single(_tiles, self, strand, kwargs)
return PyRanges(df)
def window(self, window_size, strand=None, **kwargs):
from pyranges.methods.windows import _windows
kwargs["sparse"] = {"self": False}
kwargs["window_size"] = window_size
df = pyrange_apply_single(_windows, self, strand, kwargs)
return PyRanges(df)
def apply(self, f, strand=False, as_pyranges=True, **kwargs):
kwargs.update({"strand": strand})
kwargs = fill_kwargs(kwargs)
result = pyrange_apply_single(f, self, strand, kwargs)
if not as_pyranges:
return result
else:
return PyRanges(result)
def slack(self, slack):
if isinstance(slack, dict):
assert self.stranded, "PyRanges must be stranded to add 5/3-end specific slack."
kwargs = fill_kwargs({"slack": slack})
prg = PyRanges(
pyrange_apply_single(_slack, self, self.stranded, kwargs))
return prg
def split(self, strand=None, **kwargs):
if strand is None:
strand = self.stranded
kwargs = fill_kwargs(kwargs)
from pyranges.methods.split import _split
df = pyrange_apply_single(_split, self, strand, kwargs)
return pr.PyRanges(df)
def drop_duplicate_positions(self, strand=None, **kwargs):
from pyranges.methods.drop_duplicates import _drop_duplicate_positions
if strand is None:
strand = self.stranded
kwargs["sparse"] = {"self": False}
kwargs = fill_kwargs(kwargs)
kwargs["strand"] = strand and self.stranded
return PyRanges(
pyrange_apply_single(_drop_duplicate_positions, self, strand,
kwargs))
def eval(self, eval_cmd, strand=True, as_pyranges=True, **kwargs):
f = lambda df: eval(eval_cmd)
kwargs = fill_kwargs(kwargs)
result = pyrange_apply_single(f, self, strand, kwargs)
if not as_pyranges:
return result
else:
return PyRanges(result)
def new_position(self, new_pos, strand=None, **kwargs):
from pyranges.methods.new_position import _new_position
kwargs["sparse"] = {"self": False}
kwargs["new_pos"] = new_pos
kwargs = fill_kwargs(kwargs)
if strand is None:
strand = self.stranded
dfs = pyrange_apply_single(_new_position, self, strand, kwargs)
return pr.PyRanges(dfs)
def apply(self, f, strand=False, as_pyranges=True, kwargs=None):
if not kwargs:
kwargs = {}
kwargs = fill_kwargs(kwargs)
f = ray.remote(f)
result = pyrange_apply_single(f, self, strand, kwargs)
if not as_pyranges:
return result
else:
return PyRanges(result)
def apply(self, f, strand=None, as_pyranges=True, **kwargs):
if strand is None:
strand = self.stranded
kwargs.update({"strand": strand})
kwargs.update(kwargs.get("kwargs", {}))
kwargs = fill_kwargs(kwargs)
result = pyrange_apply_single(f, self, strand, kwargs)
if not as_pyranges:
return result
else:
return PyRanges(result)
def set_columns(self, value):
assert len(value) == len(
self.columns), "New and old columns must be same length"
def _columns(df):
df.columns = value
return df
return pr.PyRanges(
pyrange_apply_single(_columns,
self,
strand=None,
kwargs={"sparse": {
"self": False
}}))
def assign(self, col, function, strand=False, **kwargs):
kwargs = fill_kwargs(kwargs)
result = pyrange_apply_single(function, self, strand, kwargs)
first_result = next(iter(result.values()))
assert type(
first_result
) == pd.Series, "result of assign function must be Series, but is {}".format(
type(first_result))
# do a deepcopy of object
new_self = pr.PyRanges({k: v.copy() for k, v in self.items()})
new_self.__setattr__(col, result)
return new_self
def _coverage(ranges, value_col=None, strand=True, rpm=False):
try:
from pyrle.methods import coverage
from pyrle import PyRles
except ImportError:
raise Exception(
"Using the coverage method requires that pyrle is installed.")
kwargs = {"value_col": value_col}
if value_col is None:
kwargs["sparse"] = {"self": True}
# from pydbg import dbg
result = pyrange_apply_single(coverage, ranges, strand, kwargs)
if rpm:
multiplier = 1e6 / len(ranges)
result = {k: v * multiplier for k, v in result.items()}
return PyRles(result)
def subset(self, function, strand=None, **kwargs):
kwargs = fill_kwargs(kwargs)
if strand is None:
strand = self.stranded
if self.stranded and not strand:
self = self.unstrand()
result = pyrange_apply_single(function, self, strand, kwargs)
if not result:
return pr.PyRanges()
first_result = next(iter(result.values()))
assert first_result.dtype == bool, "result of subset function must be bool, but is {}".format(
first_result.dtype)
return self[result]
def tesify(self, slack=0):
kwargs = fill_kwargs({"slack": slack})
return PyRanges(
pyrange_apply_single(_tes, self, self.stranded, kwargs))
def sort(self, columns=("Start", "End"), **kwargs):
from pyranges.methods.sort import _sort
kwargs["sparse"] = {"self": False}
kwargs = fill_kwargs(kwargs)
return PyRanges(
pyrange_apply_single(_sort, self, self.stranded, kwargs))
def three_end(self, slack=0):
assert self.stranded, "Need stranded pyrange to find 3'."
kwargs = fill_kwargs({"slack": slack})
return PyRanges(pyrange_apply_single(_tes, self, self.stranded,
kwargs))