-
Notifications
You must be signed in to change notification settings - Fork 2
/
interval_tree.py
129 lines (106 loc) · 4.89 KB
/
interval_tree.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
import bisect
import Sequencing.genomes as genomes
import gtf
from collections import defaultdict
class IntervalTree(object):
def __init__(self, intervals):
sorted_intervals = sorted(intervals)
# Pick a balancing center point by finding the middle of the middle
# interval in the sorted list.
middle_interval = sorted_intervals[len(sorted_intervals) / 2]
self.center_point = (middle_interval.start + middle_interval.end) / 2
to_left = []
overlapping = []
to_right = []
for interval in sorted_intervals:
if interval.end < self.center_point:
to_left.append(interval)
elif interval.start > self.center_point:
to_right.append(interval)
else:
overlapping.append(interval)
if to_left:
self.left_subtree = IntervalTree(to_left)
else:
self.left_subtree = None
if to_right:
self.right_subtree = IntervalTree(to_right)
else:
self.right_subtree = None
self.overlapping_by_start = sorted(overlapping, key=lambda interval: interval.start)
self.overlapping_starts = [interval.start for interval in self.overlapping_by_start]
self.overlapping_by_end = sorted(overlapping, key=lambda interval: interval.end)
self.overlapping_ends = [interval.end for interval in self.overlapping_by_end]
def containing_point(self, point):
if point == self.center_point:
overlapping = self.overlapping_by_start
elif point < self.center_point:
index = bisect.bisect(self.overlapping_starts, point)
overlapping = self.overlapping_by_start[:index]
if self.left_subtree:
overlapping += self.left_subtree.containing_point(point)
elif point > self.center_point:
index = bisect.bisect_left(self.overlapping_ends, point)
overlapping = self.overlapping_by_end[index:]
if self.right_subtree:
overlapping += self.right_subtree.containing_point(point)
return overlapping
class IntervalEndpointList(object):
def __init__(self, intervals):
starts = [(interval.start, interval) for interval in intervals]
ends = [(interval.end, interval) for interval in intervals]
self.endpoint_list = sorted(starts + ends)
self.endpoints = [endpoint for endpoint, interval in self.endpoint_list]
def endpoint_contained_in(self, start, end):
first_index = bisect.bisect_left(self.endpoints, start)
past_last_index = bisect.bisect_right(self.endpoints, end)
overlapping_endpoints = self.endpoint_list[first_index:past_last_index]
# Note: intervals may contain duplicates.
intervals = [interval for endpoint, interval in overlapping_endpoints]
return intervals
class OverlapFinder(object):
def __init__(self, intervals):
self.interval_tree = IntervalTree(intervals)
self.endpoint_list = IntervalEndpointList(intervals)
def overlapping(self, start, end):
from_tree = self.interval_tree.containing_point(start)
from_list = self.endpoint_list.endpoint_contained_in(start, end)
return sorted(set(from_tree + from_list))
class NamedOverlapFinder(object):
def __init__(self, named_intervals, genome_dir):
by_name = defaultdict(list)
for interval in named_intervals:
by_name[interval.seqname].append(interval)
genome_index = genomes.get_genome_index(genome_dir)
for name in by_name:
start = gtf.Feature.sequence_edge(name, -1)
end = gtf.Feature.sequence_edge(name, genome_index[name].length)
by_name[name].append(start)
by_name[name].append(end)
self.overlap_finders = {}
for name in by_name:
self.overlap_finders[name] = OverlapFinder(by_name[name])
def overlapping(self, seqname, start, end):
return self.overlap_finders[seqname].overlapping(start, end)
def find_closest_before(self, seqname, strand, start):
window = 100
before = []
while before == []:
overlapping = self.overlapping(seqname, start - window, start - 1)
before = [f for f in overlapping
if f.end < start and f.strand in ['.', strand]
]
window *= 10
before = sorted(before, key=lambda f: f.end, reverse=True)
return before
def find_closest_after(self, seqname, strand, end):
window = 100
after = []
while after == []:
overlapping = self.overlapping(seqname, end + 1, end + window)
after = [f for f in overlapping
if f.start > end and f.strand in ['.', strand]
]
window *= 10
after = sorted(after, key=lambda f: f.start)
return after