Added an extra min_length threshold check in find_ibd.

Added extra find_ibd tests using min_length input.

Author: gtsambos

c/tskit/tables.c

@@ -7415,17 +7415,19 @@ tsk_ibd_finder_enqueue_segment(
     void *p;
 
     tsk_bug_assert(left < right);
-    /* Make sure we always have room for one more segment in the queue so we
-     * can put a tail sentinel on it */
-    if (self->segment_queue_size == self->max_segment_queue_size - 1) {
-        self->max_segment_queue_size *= 2;
-        p = tsk_realloc(self->segment_queue,
-            self->max_segment_queue_size * sizeof(*self->segment_queue));
-        if (p == NULL) {
-            ret = TSK_ERR_NO_MEMORY;
-            goto out;
+    if (right - left > self->min_length) {
+        /* Make sure we always have room for one more segment in the queue so we
+         * can put a tail sentinel on it */
+        if (self->segment_queue_size == self->max_segment_queue_size - 1) {
+            self->max_segment_queue_size *= 2;
+            p = tsk_realloc(self->segment_queue,
+                self->max_segment_queue_size * sizeof(*self->segment_queue));
+            if (p == NULL) {
+                ret = TSK_ERR_NO_MEMORY;
+                goto out;
+            }
+            self->segment_queue = p;
         }
-        self->segment_queue = p;
     }
     seg = self->segment_queue + self->segment_queue_size;
     seg->left = left;

python/tests/ibd.py

@@ -266,7 +266,7 @@ def find_ibd_segments(self):
                             max(seg.left, s.left),
                             min(seg.right, s.right),
                         )
-                        if intvl[1] - intvl[0] > 0:
+                        if intvl[1] - intvl[0] > self.min_length:
                             list_to_add.add(Segment(intvl[0], intvl[1], s.node))
                         s = s.next
 

python/tests/test_ibd.py

@@ -778,6 +778,51 @@ def test_defaults(self):
         assert ibd_is_equal(ibd_segs, true_segs)
 
 
+class TestIbdLengthThreshold:
+    """
+    Tests the behaviour of the min_length argument in niche cases.
+    """
+
+    # 2
+    #             |     3
+    # 1      2    |    / \
+    #       / \   |   /   \
+    # 0    0   1  |  0     1
+    # |------------|----------|
+    # 0.0          0.4        1.0
+    nodes = io.StringIO(
+        """\
+    id      is_sample   time
+    0       1           0
+    1       1           0
+    2       0           1
+    3       0           1.5
+    """
+    )
+    edges = io.StringIO(
+        """\
+    left    right   parent  child
+    0       0.4     2       0,1
+    0.4     1.0     3       0,1
+    """
+    )
+    ts = tskit.load_text(nodes=nodes, edges=edges, strict=False)
+
+    def test_length_exceeds_segment(self):
+        ibd_segs = find_ibd(self.ts, sample_pairs=[(0, 1)], min_length=1.1)
+        true_segs = {(0, 1): []}
+        assert ibd_is_equal(ibd_segs, true_segs)
+
+    def test_length_is_negative(self):
+        with pytest.raises(tskit.LibraryError):
+            find_ibd(self.ts, sample_pairs=[(0, 1)], min_length=-0.1)
+
+    def test_equal_to_length(self):
+        ibd_segs = find_ibd(self.ts, sample_pairs=[(0, 1)], min_length=0.4)
+        true_segs = {(0, 1): [ibd.Segment(0.4, 1.0, 3)]}
+        assert ibd_is_equal(ibd_segs, true_segs)
+
+
 class TestIbdRandomExamples:
     """
     Randomly generated test cases.