def _cluster(self, strand=False, maxdist=0, minnb=1):
dfs = []
idx_start, idx_end = 0, 0
if strand:
for (c, s), cdf in self.df.groupby(["Chromosome", "Strand"]):
cdf = cdf.sort_values("Start")
starts, ends = find_clusters(cdf.Start.values, cdf.End.values)
df = pd.DataFrame({"Chromosome": c, "Start": starts, "End": ends, "Strand": s})
dfs.append(df)
df = pd.concat(dfs, ignore_index=True)["Chromosome Start End Strand".split()]
else:
for c, cdf in self.df.groupby(["Chromosome"]):
cdf = cdf.sort_values("Start")
starts, ends = find_clusters(cdf.Start.values, cdf.End.values)
df = pd.DataFrame({"Chromosome": c, "Start": starts, "End": ends})
dfs.append(df)
df = pd.concat(dfs, ignore_index=True)["Chromosome Start End".split()]
return df
def _merge(df, kwargs):
if df.empty:
return None
chromosome, strand = kwargs["chromosome"], kwargs.get("strand", None)
cdf = df.sort_values("Start")
starts, ends = find_clusters(cdf.Start.values, cdf.End.values)
nidx = pd.Index(range(len(starts)))
if strand:
cluster_df = pd.DataFrame({
"Chromosome":
pd.Series(chromosome, dtype="category", index=nidx),
"Start":
starts,
"End":
ends,
"Strand":
pd.Series(strand, dtype="category", index=nidx)
})
else:
cluster_df = pd.DataFrame({
"Chromosome":
pd.Series(chromosome, dtype="category", index=nidx),
"Start":
starts,
"End":
ends
})
return cluster_df
def _set_union(scdf, ocdf, **kwargs):
chromosome, strand = parse_grpby_key(kwargs["key"])
strandedness = kwargs["strandedness"]
strand = True if strandedness == "same" else False
if len(scdf) == 0:
return _cluster(ocdf, chromosome, strand=strand)
elif len(ocdf) == 0:
return _cluster(scdf, chromosome, strand=strand)
_starts = np.concatenate([scdf.Start.values, ocdf.Start.values])
_ends = np.concatenate([scdf.End.values, ocdf.End.values])
cdf = pd.DataFrame({"Start": _starts, "End": _ends})["Start End".split()]
cdf = cdf.sort_values("Start")
starts, ends = find_clusters(cdf.Start.values, cdf.End.values)
chromosome = scdf.head(1)["Chromosome"].iloc[0]
if strandedness == "same":
_strand = scdf.head(1)["Strand"].iloc[0]
else:
_strand = False
cluster_df = _create_df_from_starts_ends(starts, ends, chromosome, _strand)
return cluster_df
def _cluster(df, chromosome, strand=False, **kwargs):
cdf = df.sort_values("Start")
starts, ends = find_clusters(cdf.Start.values, cdf.End.values)
cluster_df = _create_df_from_starts_ends(starts, ends, chromosome, strand)
return cluster_df
def _merge(df, **kwargs):
if df.empty:
return None
slack = kwargs.get("slack", 0)
chromosome, strand = kwargs["chromosome"], kwargs.get("strand", None)
cdf = df.sort_values("Start")
starts, ends, number = find_clusters(cdf.Start.values, cdf.End.values,
slack)
nidx = pd.Index(range(len(starts)))
if strand:
cluster_df = pd.DataFrame({
"Chromosome":
pd.Series(chromosome, dtype="category", index=nidx),
"Start":
starts,
"End":
ends,
"Strand":
pd.Series(strand, dtype="category", index=nidx),
})
else:
cluster_df = pd.DataFrame({
"Chromosome":
pd.Series(chromosome, dtype="category", index=nidx),
"Start":
starts,
"End":
ends
})
if kwargs["count"]:
cluster_df.insert(cluster_df.shape[1], kwargs["count_col"], number)
return cluster_df
def _set_union(self, other, strand):
if strand:
assert self.stranded and other.stranded, \
"Can only do stranded set union when both PyRanges contain strand info."
if len(self) == 0:
return _cluster(other, strand=strand)
elif len(other) == 0:
return _cluster(self, strand=strand)
dfs = []
if strand and len(list(self.__ncls__.keys())[0]) == 2: # chromosome and strand
grpby_key = "Chromosome Strand".split()
columns = "Chromosome Start End Strand".split()
else:
grpby_key = "Chromosome"
columns = "Chromosome Start End".split()
self_dfs = {k: d for k, d in self.df.groupby(grpby_key)}
other_dfs = {k: d for k, d in other.df.groupby(grpby_key)}
idx_start, idx_end = 0, 0
for key in set(self_dfs).union(other_dfs):
if key in other_dfs and key in self_dfs:
_starts = np.concatenate([
self_dfs[key].Start.values,
other_dfs[key].Start.values])
_ends = np.concatenate([
self_dfs[key].End.values,
other_dfs[key].End.values])
elif key in self_dfs and not key in other_dfs:
_starts = self_dfs[key].Start.values
_ends = self_dfs[key].End.values
elif key in other_dfs and not key in self_dfs:
_starts = other_dfs[key].Start.values
_ends = other_dfs[key].End.values
cdf = pd.DataFrame({"Start": _starts, "End": _ends})["Start End".split()]
# print("cdf" * 10 + "\n", cdf)
# cdf = sort_one_by_one(cdf, "Start", "End")
cdf = cdf.sort_values("Start")
# print("cdf" * 10 + "\n", cdf)
# print(clusters)
starts, ends = find_clusters(cdf.Start.values, cdf.End.values)
# print(starts, ends)
idx_end += len(starts)
if strand:
_df = pd.DataFrame({"Chromosome": key[0], "Start": starts, "End": ends, "Strand": key[1]})
else:
_df = pd.DataFrame({"Chromosome": key, "Start": starts, "End": ends})
_df.index = range(idx_start, idx_end)
dfs.append(_df)
idx_start = idx_end
df = pd.concat(dfs)[columns]
return df