Merge pull request #352 from jeromekelleher/allow-location-nans

Allow location nans

Author: jeromekelleher

c/tests/test_trees.c

@@ -1849,17 +1849,15 @@ test_simplest_individuals(void)
     CU_ASSERT_EQUAL_FATAL(ret, TSK_ERR_INDIVIDUAL_OUT_OF_BOUNDS);
     tsk_treeseq_free(&ts);
 
+    /* NaN/ifinity values are allowed in locations they do not
+     * affect the integrity of the model. */
     tables.individuals.location[0] = NAN;
     ret = tsk_treeseq_init(&ts, &tables, load_flags);
-    CU_ASSERT_EQUAL(ret, TSK_ERR_SPATIAL_LOCATION_NONFINITE);
-    tsk_treeseq_free(&ts);
-    tables.individuals.location[0] = 0.25;
-
-    tables.individuals.location[2] = INFINITY;
-    ret = tsk_treeseq_init(&ts, &tables, load_flags);
-    CU_ASSERT_EQUAL(ret, TSK_ERR_SPATIAL_LOCATION_NONFINITE);
+    CU_ASSERT_EQUAL(ret, 0);
+    ret = tsk_treeseq_get_individual(&ts, 0, &individual);
+    CU_ASSERT_EQUAL_FATAL(ret, 0);
+    CU_ASSERT(! isfinite(individual.location[0]));
     tsk_treeseq_free(&ts);
-    tables.individuals.location[0] = 0.25;
 
     tsk_table_collection_free(&tables);
 }

c/tskit/core.c

@@ -204,9 +204,6 @@ tsk_strerror_internal(int err)
         case TSK_ERR_GENOME_COORDS_NONFINITE:
             ret = "Genome coordinates must be finite numbers";
             break;
-        case TSK_ERR_SPATIAL_LOCATION_NONFINITE:
-            ret = "Location values must be finite numbers";
-            break;
 
         /* Edge errors */
         case TSK_ERR_NULL_PARENT:

c/tskit/core.h

@@ -156,7 +156,6 @@ of tskit.
 #define TSK_ERR_PROVENANCE_OUT_OF_BOUNDS                            -209
 #define TSK_ERR_TIME_NONFINITE                                      -210
 #define TSK_ERR_GENOME_COORDS_NONFINITE                             -211
-#define TSK_ERR_SPATIAL_LOCATION_NONFINITE                            -212
 
 /* Edge errors */
 #define TSK_ERR_NULL_PARENT                                         -300

c/tskit/tables.c

@@ -5672,14 +5672,6 @@ tsk_table_collection_check_integrity(tsk_table_collection_t *self, tsk_flags_t o
         goto out;
     }
 
-    /* Individuals */
-    for (j = 0; j < self->individuals.location_offset[self->individuals.num_rows]; j++) {
-        if (! isfinite(self->individuals.location[j])) {
-            ret = TSK_ERR_SPATIAL_LOCATION_NONFINITE;
-            goto out;
-        }
-    }
-
     /* Nodes */
     for (j = 0; j < self->nodes.num_rows; j++) {
         node_time = self->nodes.time[j];

python/CHANGELOG.rst

@@ -1,12 +1,16 @@
 --------------------
-[0.2.2] - xxxx-xx-xx
+[0.2.2] - 2019-09-01
 --------------------
 
-In development.
+Minor bugfix release.
+
+Relaxes overly-strict input requirements on individual location data that
+caused some SLiM tree sequences to fail loading in version 0.2.1
+(see :issue:`351`).
 
 **New features**
 
-- Add log_time height scaling option for drawing SVG trees 
+- Add log_time height scaling option for drawing SVG trees
   (:user:`marianne-aspbury`). See :pr:`324` and :issue:`303`.
 
 **Bugfixes**

python/tests/test_tables.py

@@ -1396,45 +1396,68 @@ def test_round_trip(self):
         self.assertTrue(
             all(a == b for a, b in zip(a.tables, b.tables) if a[0] != 'provenances'))
 
-    def test_bad_edge_coords(self):
-        ts = msprime.simulate(5,  mutation_rate=1, random_seed=self.random_seed)
+
+class TestNanDoubleValues(unittest.TestCase):
+    """
+    In some tables we need to guard against NaN/infinite values in the input.
+    """
+
+    def test_edge_coords(self):
+        ts = msprime.simulate(5,  mutation_rate=1, random_seed=42)
 
         tables = ts.dump_tables()
         bad_coords = tables.edges.left + float('inf')
-        tables.edges.set_columns(**dict(tables.edges.asdict(), left=bad_coords))
+        tables.edges.left = bad_coords
         self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
 
         tables = ts.dump_tables()
         bad_coords = tables.edges.right + float('nan')
-        tables.edges.set_columns(**dict(tables.edges.asdict(), right=bad_coords))
+        tables.edges.right = bad_coords
+        self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
+
+    def test_migrations(self):
+        ts = msprime.simulate(5,  mutation_rate=1, random_seed=42)
+
+        tables = ts.dump_tables()
+        tables.populations.add_row()
+        tables.migrations.add_row(float('inf'), 1, time=0, node=0, source=0, dest=1)
         self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
 
-    def test_bad_site_positions(self):
-        ts = msprime.simulate(5,  mutation_rate=1, random_seed=self.random_seed)
+        tables = ts.dump_tables()
+        tables.populations.add_row()
+        tables.migrations.add_row(0, float('nan'), time=0, node=0, source=0, dest=1)
+        self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
+
+        tables = ts.dump_tables()
+        tables.populations.add_row()
+        tables.migrations.add_row(0, 1, time=float('nan'), node=0, source=0, dest=1)
+        self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
+
+    def test_site_positions(self):
+        ts = msprime.simulate(5,  mutation_rate=1, random_seed=42)
         tables = ts.dump_tables()
         bad_pos = tables.sites.position.copy()
         bad_pos[-1] = np.inf
-        tables.sites.set_columns(**dict(tables.sites.asdict(), position=bad_pos))
+        tables.sites.position = bad_pos
         self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
 
-    def test_bad_node_times(self):
-        ts = msprime.simulate(5,  mutation_rate=1, random_seed=self.random_seed)
+    def test_node_times(self):
+        ts = msprime.simulate(5,  mutation_rate=1, random_seed=42)
         tables = ts.dump_tables()
         bad_times = tables.nodes.time.copy()
         bad_times[-1] = np.inf
-        tables.nodes.set_columns(**dict(tables.nodes.asdict(), time=bad_times))
+        tables.nodes.time = bad_times
         self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
 
-    def test_bad_individual(self):
-        ts = msprime.simulate(12,  mutation_rate=1, random_seed=self.random_seed)
-        ts = tsutil.insert_random_ploidy_individuals(ts, seed=self.random_seed)
+    def test_individual(self):
+        ts = msprime.simulate(12,  mutation_rate=1, random_seed=42)
+        ts = tsutil.insert_random_ploidy_individuals(ts, seed=42)
         self.assertGreater(ts.num_individuals, 1)
         tables = ts.dump_tables()
         bad_locations = tables.individuals.location.copy()
         bad_locations[0] = np.inf
-        tables.individuals.set_columns(
-            **dict(tables.individuals.asdict(), location=bad_locations))
-        self.assertRaises(_tskit.LibraryError, tables.tree_sequence)
+        tables.individuals.location = bad_locations
+        ts = tables.tree_sequence()
 
 
 class TestSimplifyTables(unittest.TestCase):

python/tskit/_version.py

@@ -1,3 +1,3 @@
 # Definitive location for the version number.
 # During development, should be x.y.z.devN
-tskit_version = "0.2.2.dev0"
+tskit_version = "0.2.2"