tuning

src/scripts/cluster.py

@@ -187,7 +187,7 @@ def run_clustering (graph_gfa, mashmap_sim, hic_byread, output_dir, no_rdna, une
     LOGGING_FILENAME = "hicverkko.log"
 
     CLEAR_HOMOLOGY = 500000 #for clear bulge like structures this limmit is decreased
-    MIN_ALIGNMENT = 100000 #smaller alingments will be filtered out
+    MIN_ALIGNMENT = 50000 #smaller alingments will be filtered out
     MAX_COV = 100  # tempora# ry coverage cutoff, currently replaced by median coverage from gfa
     FIXED_WEIGHT = 100000  # best result so far with 100000 #currently replaced with max pairwise weight among datasets
 

src/scripts/scaffolding/match_graph.py

@@ -177,7 +177,7 @@ class MatchGraph:
     CLEAR_BEST = 1.5
 
     #homologous intervals should cover at least 1/3 of at least one of the nodes in pair
-    REQUIRED_COVERAGE_FRACTION = 1/3
+    REQUIRED_COVERAGE_FRACTION = 1/10
 
     def __init__(self, mashmap_sim, G, homology_weight, min_big_homology, min_alignment, logger):
         self.matchGraph = nx.Graph()
@@ -224,7 +224,7 @@ def __init__(self, mashmap_sim, G, homology_weight, min_big_homology, min_alignm
             #homology storage may be asymetric, mashmap do not guararntee anything        
             #possibly should forcely symmetrize... 
             if self.hom_storage.isValid(ec[0], ec[1]) and (not G.has_edge(ec[0], ec[1])):
-                long_enough = True
+                long_enough = False
                 for i in range (0, 2):
                     best_homology = True
                     best_len = self.matchGraph.edges[ec]['homology_len']
@@ -233,8 +233,8 @@ def __init__(self, mashmap_sim, G, homology_weight, min_big_homology, min_alignm
                             best_homology = False
                     clear_best_match = clear_best_match or best_homology
                     #we have total length of homologous sequences without joining and approximate positions with joining, check if any is long enough
-                    if max(best_len, self.hom_storage.getApproximatePositionLength(ec[0], ec[1], i)) < self.hom_storage.getLength(ec[i]) * MatchGraph.REQUIRED_COVERAGE_FRACTION:
-                        long_enough = False
+                    if max(best_len, self.hom_storage.getApproximatePositionLength(ec[0], ec[1], i)) > self.hom_storage.getLength(ec[i]) * MatchGraph.REQUIRED_COVERAGE_FRACTION:
+                        long_enough = True
                 if clear_best_match and long_enough:
                     self.matchGraph.add_edge(ec[0], ec[1], weight = homology_weight, homology_len = best_len, intervals = self.hom_storage.homologies[ec[0]][ec[1]].filtered_intervals, 
                                         orientation = self.hom_storage.homologies[ec[0]][ec[1]].orientation)