def test_create_from_nongraphpeakcollection(self):
graph = Graph({
1: Block(10),
2: Block(10),
3: Block(10)
}, {
1: [2],
2: [3]
})
graph.convert_to_numpy_backend()
linear_path = Interval(0, 10, [1, 2, 3], graph)
linear_path = linear_path.to_numpy_indexed_interval()
nongraph_peaks = NonGraphPeakCollection([
NonGraphPeak("chr1", 3, 10, 5),
NonGraphPeak("chr1", 13, 15, 7),
])
peaks = PeakCollection.create_from_nongraph_peak_collection(
graph, nongraph_peaks, linear_path, None)
self.assertEqual(peaks.intervals[0], Interval(3, 10, [1]))
self.assertEqual(peaks.intervals[1], Interval(3, 5, [2]))
peaks = PeakCollection.create_from_nongraph_peak_collection(
graph, nongraph_peaks, linear_path, LinearRegion("chr1", 3, 20))
self.assertEqual(peaks.intervals[0], Interval(0, 7, [1]))
self.assertEqual(peaks.intervals[1], Interval(0, 2, [2]))
def test_multiple_peaks3(self):
self.peaks = [
DirectedInterval(2, 2, [1, 3, 4], self.graph),
DirectedInterval(0, 3, [7, 9], self.graph)
]
self.do_asserts()
def test_multiple_peaks(self):
self.peaks = [
DirectedInterval(3, 8, [1], self.graph),
DirectedInterval(7, 12, [4], self.graph),
DirectedInterval(14, 4, [3, 4], self.graph),
]
self.do_asserts()
def test_multiple_peaks2(self):
self.peaks = [
DirectedInterval(2, 2, [1, 3, 4], self.graph),
DirectedInterval(2, 2, [6, 10, 11], self.graph),
]
self.do_asserts()
def test_extend_cyclic(self):
graph = get_small_cyclic_graph()
interval = DirectedInterval(70, 90, [1], graph=graph)
extender = Extender(graph, 40)
areas = extender.extend_interval(interval)
true_areas = BinaryContinousAreas(graph)
true_areas.add_start(1, 10)
true_areas.add_start(-1, 30)
self.assertEqual(areas, true_areas)
def test_extend_large_cyclic(self):
logging.debug("CYCLIC")
graph = get_large_cyclic_graph()
interval = DirectedInterval(90, 10, [1, 2], graph=graph)
extender = Extender(graph, 40)
areas = extender.extend_interval(interval)
true_areas = BinaryContinousAreas(graph)
true_areas.add_start(1, 10)
true_areas.add_start(-1, 10)
true_areas.add_start(2, 20)
self.assertEqual(areas, true_areas)
def test_convert_to_approx_linear_peaks(self):
graph = Graph({i: Block(3)
for i in range(1, 10)}, {
1: [2],
2: [3],
3: [4],
4: [5],
5: [6],
6: [7, 8],
7: [9],
9: [9]
})
graph.convert_to_numpy_backend()
linear_interval = Interval(0, 3, [2, 4, 8, 9], graph)
linear_interval = linear_interval.to_numpy_indexed_interval()
peaks = PeakCollection([Peak(2, 2, [2, 3, 4]), Peak(1, 1, [3, 4, 5])])
linear_peaks = peaks.to_approx_linear_peaks(linear_interval, "chr4")
linear_peaks = linear_peaks.peaks
print(linear_peaks)
self.assertEqual(linear_peaks[0], NonGraphPeak("chr4", 2, 5))
self.assertEqual(linear_peaks[1], NonGraphPeak("chr4", 3, 3))
def _get_graph_interval(self, tmp_start, tmp_end, direction):
start = tmp_start
end = tmp_end
if direction == -1:
start = -tmp_end
end = -tmp_start
start_rp = start // self.node_size
end_rp = (end + 1) // self.node_size
region_paths = list(range(start_rp, end_rp))
start_pos = Position(start_rp, start % self.node_size)
end_pos = Position(end_rp, (end % self.node_size) + 1)
return DirectedInterval(start_pos,
end_pos,
region_paths,
direction=direction)
def neg_linear_to_graph_interval(self, lin_interval):
start_rp = ((lin_interval.end - 1) // self.node_size + 1)
end_rp = (lin_interval.start // self.node_size + 1)
start_offset = start_rp * self.node_size - lin_interval.end
end_offset = end_rp * self.node_size - lin_interval.start
# start_offset = (-lin_interval.end) % self.node_size
# end_offset = (-lin_interval.start) % self.node_size
# start_rp = (lin_interval.end) // self.node_size + 1
# end_rp = (-lin_interval.start) // self.node_size + 1
rps = list(range(start_rp * -1, end_rp * -1 + 1))
interval = DirectedInterval(start_offset,
end_offset,
rps,
graph=self.graph)
return interval
def linear_to_graph_interval(self, lin_interval, is_control=None):
if lin_interval.direction == -1:
return self.neg_linear_to_graph_interval(lin_interval)
start = lin_interval.start
end = lin_interval.end
start_rp = start // self.node_size + 1
end_rp = (end - 1) // self.node_size + 1
start_pos = Position(start_rp, start % self.node_size)
end_pos = Position(end_rp, ((end - 1) % self.node_size) + 1)
region_paths = list(range(start_rp, end_rp + 1))
interval = DirectedInterval(start_pos,
end_pos,
region_paths,
direction=lin_interval.direction,
graph=self.graph)
return interval
def _create_reads(self, chrom_number, chrom, graph):
i = chrom_number
sample_reads = []
control_reads = []
peaks = [DirectedInterval(7, 2, [1 + 3 * i, 2 + 3 * i], graph)]
self.peaks.append(peaks)
for peak in peaks:
for i in range(0, 10):
left_sub = peak.get_subinterval(0, self.read_length)
sample_reads.append(left_sub)
control_reads.append(left_sub)
right_sub = peak.get_subinterval(
self.fragment_length - self.read_length,
self.fragment_length)
right_sub_reverse = right_sub.get_reverse()
sample_reads.append(right_sub_reverse)
control_reads.append(right_sub_reverse)
self.sample_reads.append(Intervals(sample_reads))
self.control_reads.append(Intervals(control_reads))
def assertPeakSetsEqual(self, linear_peaks_file, graph_peaks_file):
linear_path = DirectedInterval(0,
self.node_size,
list(range(1, self.n_nodes + 1)),
graph=self.graph)
comparer = PeaksComparer.create_from_graph_peaks_and_linear_peaks(
linear_peaks_file,
graph_peaks_file,
self.graph,
linear_path,
graph_region=None)
# for i, j in zip(sorted(comparer.peaks1.intervals, key=lambda x: x.region_paths[0]),
# sorted(comparer.peaks2.intervals, key=lambda x: x.region_paths[0])):
# print(i, j)
assert len(comparer.peaks1.intervals) == len(comparer.peaks2.intervals)
matches = comparer.get_peaks_at_same_position()
# for m in matches:
# print(m)
assert len(matches) == len(comparer.peaks1.intervals)
def test_peak_crossing_blocks(self):
self.peaks = [
DirectedInterval(13, 3, [1, 2], self.graph),
DirectedInterval(13, 3, [2, 4], self.graph),
]
self.do_asserts()
def test_single_peak(self):
self.peaks = [
DirectedInterval(3, 8, [2], self.graph),
]
self.do_asserts()
def test_single_peak(self):
self.peaks = [
DirectedInterval(1, 1, [1, 2, 7], self.graph),
]
self.do_asserts()
if pytest.__version__ < "3.0.0":
pytest.skip()
# from graph_peak_caller.extender import Extender, Areas
# from graph_peak_caller.areas import BinaryContinousAreas
from offsetbasedgraph import Block, GraphWithReversals,\
DirectedInterval, Position
from cyclic_graph import get_small_cyclic_graph, get_large_cyclic_graph
logging.getLogger("extender").setLevel("DEBUG")
Graph = GraphWithReversals
nodes = {i: Block(20) for i in range(1, 4)}
tmp_edges = {1: [2, -2], 2: [3], -2: [3]}
graph = Graph(nodes, tmp_edges)
pos_interval = DirectedInterval(8, 18, [2])
neg_interval = DirectedInterval(2, 12, [-2])
neg_spanning_interval = DirectedInterval(12, 2, [-3, -2])
pos_spanning_interval = DirectedInterval(18, 8, [2, 3])
@pytest.mark.skip("Legacy")
class TestExtender(unittest.TestCase):
def setUp(self):
self.extender = Extender(graph, 20)
def test_extend(self):
areas = self.extender.extend_interval(pos_interval)
true_areas = BinaryContinousAreas(graph)
true_areas.add_start(-2, 12)
true_areas.add_start(3, 8)
