Merge pull request #19 from bcgsc/release/v1.5.1

Release v1.5.1: Summary split read call fix

mavis/cluster/constants.py

@@ -24,7 +24,7 @@
     'max_proximity', 5000,
     defn='the maximum distance away from an annotation before the region in considered to be uninformative')
 DEFAULTS.add(
-    'uninformative_filter', True,
+    'uninformative_filter', False,
     defn='flag that determines if breakpoint pairs which are not within max_proximity to any annotations are filtered '
     'out prior to clustering')
 DEFAULTS.add(

mavis/pairing/pairing.py

@@ -165,6 +165,7 @@ def pair_by_distance(calls, distances, log=devnull, against_self=False):
     comparisons = 0
     for i in range(0, len(break1_sorted)):
         current = break1_sorted[i]
+        distance_pairings.setdefault(product_key(current), set())
         for j in range(i + 1, len(break1_sorted)):
             other = break1_sorted[j]
 
@@ -175,7 +176,7 @@ def pair_by_distance(calls, distances, log=devnull, against_self=False):
             if not against_self and current.library == other.library and current.protocol == other.protocol:
                 continue  # do not pair within a single library
             if equivalent(current, other, distances=distances):
-                distance_pairings.setdefault(product_key(current), set()).add(product_key(other))
+                distance_pairings[product_key(current)].add(product_key(other))
                 distance_pairings.setdefault(product_key(other), set()).add(product_key(current))
         current = break2_sorted[i]
         for j in range(i + 1, len(break2_sorted)):

mavis/summary/main.py

@@ -89,6 +89,7 @@ def main(
             COLUMNS.protein_synon,
             COLUMNS.cdna_synon,
             COLUMNS.net_size,
+            COLUMNS.tracking_id,
             'dgv',
             'summary_pairing']
         }, COLUMNS.call_method: CALL_METHOD.INPUT},
@@ -255,6 +256,10 @@ def main(
             annotate_dgv(bpps_by_library[lib], dgv_annotation, distance=10)  # TODO make distance a parameter
         log('adding pairing states for', lib)
         for row in bpps_by_library[lib]:
+            # in case no pairing was done, add default (applicable to single library summaries)
+            row.data.setdefault(COLUMNS.inferred_pairing, '')
+            row.data.setdefault(COLUMNS.pairing, '')
+            row.data.setdefault(COLUMNS.library, lib)
             # filter pairing ids based on what is still kept?
             paired_libraries = set([p.split('_')[0] for p in row.pairing.split(';')])
             inferred_paired_libraries = set([p.split('_')[0] for p in row.inferred_pairing.split(';')])

mavis/summary/summary.py

@@ -107,7 +107,8 @@ def group_events(events):
         COLUMNS.spanning_reads, COLUMNS.flanking_pairs, COLUMNS.tools,
         COLUMNS.product_id, COLUMNS.event_type, COLUMNS.annotation_id,
         COLUMNS.pairing, COLUMNS.annotation_figure,
-        COLUMNS.contig_remapped_reads, COLUMNS.tools
+        COLUMNS.contig_remapped_reads, COLUMNS.tools,
+        COLUMNS.tracking_id
     ]:
         new_data = sorted(list({bpp.data[col] for bpp in events}))
         new_bpp.data[col] = new_data[0] if len(new_data) == 1 else ';'.join([str(v) for v in new_data])
@@ -131,7 +132,7 @@ def group_by_distance(calls, distances):
     grouped_calls = []
     for component in get_connected_components(pairing):
         if len(component) == 1:
-            grouped_calls.append(mapping[component.pop()])
+            grouped_calls.extend(mapping[component.pop()])
         else:
             pairs = []
             for key in component:
@@ -239,15 +240,15 @@ def filter_by_evidence(
                 bpp.break2_split_reads < filter_min_split_reads,
                 all([
                     bpp.event_type != SVTYPE.INS,
-                    bpp.break2_split_reads_forced + bpp.break1_split_reads_forced < filter_min_linking_split_reads
+                    bpp.linking_split_reads < filter_min_linking_split_reads
                 ]),
                 all([
                     bpp.event_type == SVTYPE.INS,
                     bpp.flanking_pairs < filter_min_linking_split_reads,
                     bpp.break2_split_reads_forced + bpp.break1_split_reads_forced < filter_min_linking_split_reads
                 ]),
-                bpp.break1_split_reads + bpp.break2_split_reads -
-                (bpp.break2_split_reads_forced + bpp.break1_split_reads_forced) < 1
+                bpp.break1_split_reads - bpp.break1_split_reads_forced < 1,
+                bpp.break2_split_reads - bpp.break2_split_reads_forced < 1
             ]):
                 removed.append(bpp)
                 continue

setup.py

@@ -28,7 +28,7 @@ def check_nonpython_dependencies():
 
 setup(
     name='mavis',
-    version='1.5.0',
+    version='1.5.1',
     url='https://github.com/bcgsc/mavis.git',
     packages=find_packages(),
     install_requires=[