def overlap_2d_bedtools(target, reference, margin, return_ref=False):
l_target = target[['chrom1',
'pos1']].rename(columns=lambda x: x.replace('1', ''))
l_target = insert_margin(l_target, margin)
l_ref = reference[['chrom1',
'pos1']].rename(columns=lambda x: x.replace('1', ''))
l_ref = insert_margin(l_ref, margin)
with tsv(l_ref) as a, tsv(l_target) as b:
l_intersect = bedtools.intersect(a=a.name, b=b.name, wa=True, wb=True)
l_intersect.columns = [col + '_r' for col in l_ref.columns
] + [col + '_t' for col in l_target.columns]
l_intersect.set_index(['dot_id_r', 'dot_id_t'], inplace=True)
r_target = target[['chrom2',
'pos2']].rename(columns=lambda x: x.replace('2', ''))
r_target = insert_margin(r_target, margin)
r_ref = reference[['chrom2',
'pos2']].rename(columns=lambda x: x.replace('2', ''))
r_ref = insert_margin(r_ref, margin)
with tsv(r_ref) as a, tsv(r_target) as b:
r_intersect = bedtools.intersect(a=a.name, b=b.name, wa=True, wb=True)
r_intersect.columns = [col + '_r' for col in r_ref.columns
] + [col + '_t' for col in r_target.columns]
r_intersect.set_index(['dot_id_r', 'dot_id_t'], inplace=True)
merged_df = l_intersect.merge(r_intersect,
how='inner',
left_index=True,
right_index=True).reset_index()
target_inds = merged_df.dot_id_t.values
target_result = target.loc[target_inds].copy().sort_index(
).drop_duplicates()
if return_ref:
ref_inds = merged_df.dot_id_r.values
reference_result = reference.loc[ref_inds].copy().sort_index(
).drop_duplicates()
return target_result, reference_result
return target_result
def pair_sites(sites, separation, slop):
"""
Create "hand" intervals to the right and to the left of each site.
Then join right hands with left hands to pair sites together.
"""
from bioframe.tools import tsv, bedtools
mids = (sites["start"] + sites["end"]) // 2
left_hand = sites[["chrom"]].copy()
left_hand["start"] = mids - separation - slop
left_hand["end"] = mids - separation + slop
left_hand["site_id"] = left_hand.index
left_hand["direction"] = "L"
left_hand["snip_mid"] = mids
left_hand["snip_strand"] = sites["strand"]
right_hand = sites[["chrom"]].copy()
right_hand["start"] = mids + separation - slop
right_hand["end"] = mids + separation + slop
right_hand["site_id"] = right_hand.index
right_hand["direction"] = "R"
right_hand["snip_mid"] = mids
right_hand["snip_strand"] = sites["strand"]
# ignore out-of-bounds hands
mask = (left_hand["start"] > 0) & (right_hand["start"] > 0)
left_hand = left_hand[mask].copy()
right_hand = right_hand[mask].copy()
# intersect right hands (left anchor site)
# with left hands (right anchor site)
with tsv(right_hand) as R, tsv(left_hand) as L:
out = bedtools.intersect(a=R.name, b=L.name, wa=True, wb=True)
out.columns = [c + "_r" for c in right_hand.columns] + [
c + "_l" for c in left_hand.columns
]
return out
def pair_sites(sites, separation, slop):
"""
Create "hand" intervals to the right and to the left of each site.
Then join right hands with left hands to pair sites together.
"""
from bioframe.tools import tsv, bedtools
mids = (sites['start'] + sites['end']) // 2
left_hand = sites[['chrom']].copy()
left_hand['start'] = mids - separation - slop
left_hand['end'] = mids - separation + slop
left_hand['site_id'] = left_hand.index
left_hand['direction'] = 'L'
left_hand['snip_mid'] = mids
left_hand['snip_strand'] = sites['strand']
right_hand = sites[['chrom']].copy()
right_hand['start'] = mids + separation - slop
right_hand['end'] = mids + separation + slop
right_hand['site_id'] = right_hand.index
right_hand['direction'] = 'R'
right_hand['snip_mid'] = mids
right_hand['snip_strand'] = sites['strand']
# ignore out-of-bounds hands
mask = (left_hand['start'] > 0) & (right_hand['start'] > 0)
left_hand = left_hand[mask].copy()
right_hand = right_hand[mask].copy()
# intersect right hands (left anchor site)
# with left hands (right anchor site)
with tsv(right_hand) as R, tsv(left_hand) as L:
out = bedtools.intersect(a=R.name, b=L.name, wa=True, wb=True)
out.columns = ([c + '_r' for c in right_hand.columns] +
[c + '_l' for c in left_hand.columns])
return out