Fixed merge None type error. Added wb3 as default compression level for long-reads (run-pipeline) (#116)

Author: kcleal

dysgu/call_component.pyx

@@ -833,6 +833,7 @@ def linear_scan_clustering(spanning, informative):
 #     return result
 
 
+
 def process_spanning(paired_end, spanning_alignments, divergence, length_extend, informative,
                      generic_insertions, insert_ppf, to_assemble):
     # echo("PROCESS SPANNING")

dysgu/main.py

@@ -60,7 +60,8 @@
                         "trust_ins_len": "False",
                         "sd": 0.6,
                         "symbolic_sv_size": 50000,
-                        "divergence": "auto"
+                        "divergence": "auto",
+                        "compression": "wb3",
                         },
             "nanopore-r10": {"mq": 1,
                         "min_support": "auto",
@@ -72,7 +73,8 @@
                         "trust_ins_len": "False",
                         "sd": 0.35,
                         "thresholds": "0.35,0.35,0.35,0.35,0.35",
-                        "symbolic_sv_size": 50000
+                        "symbolic_sv_size": 50000,
+                        "compression": "wb3",
                         },
            "pacbio-sequel2": {"mq": 1,
                       "min_support": "auto",
@@ -83,7 +85,8 @@
                       "clip_length": -1,
                       "trust_ins_len": "True",
                       "sd": 0.45,
-                      "symbolic_sv_size": 50000
+                      "symbolic_sv_size": 50000,
+                      "compression": "wb3",
                       },
            "pacbio-revio": {"mq": 1,
                       "min_support": "auto",
@@ -95,7 +98,8 @@
                       "trust_ins_len": "True",
                       "sd": 0.4,
                       "thresholds": "0.25,0.25,0.25,0.25,0.25",
-                      "symbolic_sv_size": 50000
+                      "symbolic_sv_size": 50000,
+                      "compression": "wb3",
                       },
            "pe": {"mq": defaults["mq"],
                   "min_support": defaults["min_support"],
@@ -211,12 +215,11 @@ def cli():
 @click.option("-p", "--procs", help="Number of cpu cores to use", type=cpu_range, default=1,
               show_default=True)
 @click.option('--mode', help=f"Type of input reads. Multiple options are set, overrides other options. "
-                             f"pacbio-sequel2: --mq {presets['pacbio-sequel2']['mq']} --paired False --min-support '{presets['pacbio-sequel2']['min_support']}' --max-cov {presets['pacbio-sequel2']['max_cov']} --dist-norm {presets['pacbio-sequel2']['dist_norm']} --trust-ins-len True --symbolic-sv-size {presets['pacbio-sequel2']['symbolic_sv_size']} --sd {presets['pacbio-sequel2']['sd']}."
-                             f"pacbio-revio: --mq {presets['pacbio-revio']['mq']} --paired False --min-support '{presets['pacbio-revio']['min_support']}' --max-cov {presets['pacbio-revio']['max_cov']} --dist-norm {presets['pacbio-revio']['dist_norm']} --trust-ins-len True --thresholds {presets['pacbio-revio']['thresholds']} --symbolic-sv-size {presets['pacbio-revio']['symbolic_sv_size']} --sd {presets['pacbio-revio']['sd']}."
-                             f"nanopore-r9: --mq {presets['nanopore-r9']['mq']} --paired False --min-support '{presets['nanopore-r9']['min_support']}' --max-cov {presets['nanopore-r9']['max_cov']} --dist-norm {presets['nanopore-r9']['dist_norm']} --trust-ins-len False --symbolic-sv-size {presets['nanopore-r9']['symbolic_sv_size']} --sd {presets['nanopore-r9']['sd']} --divergence {presets['nanopore-r9']['divergence']}."
-                             f"nanopore-r10: --mq {presets['nanopore-r10']['mq']} --paired False --min-support '{presets['nanopore-r10']['min_support']}' --max-cov {presets['nanopore-r10']['max_cov']} --dist-norm {presets['nanopore-r10']['dist_norm']} --trust-ins-len False --thresholds {presets['nanopore-r10']['thresholds']} --symbolic-sv-size {presets['nanopore-r10']['symbolic_sv_size']} --sd {presets['nanopore-r10']['sd']}",
+                             f"pacbio-sequel2: --mq {presets['pacbio-sequel2']['mq']} --paired False --min-support '{presets['pacbio-sequel2']['min_support']}' --max-cov {presets['pacbio-sequel2']['max_cov']} --dist-norm {presets['pacbio-sequel2']['dist_norm']} --trust-ins-len True --symbolic-sv-size {presets['pacbio-sequel2']['symbolic_sv_size']} --sd {presets['pacbio-sequel2']['sd']} --compression wb3. "
+                             f"pacbio-revio: --mq {presets['pacbio-revio']['mq']} --paired False --min-support '{presets['pacbio-revio']['min_support']}' --max-cov {presets['pacbio-revio']['max_cov']} --dist-norm {presets['pacbio-revio']['dist_norm']} --trust-ins-len True --thresholds {presets['pacbio-revio']['thresholds']} --symbolic-sv-size {presets['pacbio-revio']['symbolic_sv_size']} --sd {presets['pacbio-revio']['sd']} --compression wb3. "
+                             f"nanopore-r9: --mq {presets['nanopore-r9']['mq']} --paired False --min-support '{presets['nanopore-r9']['min_support']}' --max-cov {presets['nanopore-r9']['max_cov']} --dist-norm {presets['nanopore-r9']['dist_norm']} --trust-ins-len False --symbolic-sv-size {presets['nanopore-r9']['symbolic_sv_size']} --sd {presets['nanopore-r9']['sd']} --divergence {presets['nanopore-r9']['divergence']} --compression wb3. "
+                             f"nanopore-r10: --mq {presets['nanopore-r10']['mq']} --paired False --min-support '{presets['nanopore-r10']['min_support']}' --max-cov {presets['nanopore-r10']['max_cov']} --dist-norm {presets['nanopore-r10']['dist_norm']} --trust-ins-len False --thresholds {presets['nanopore-r10']['thresholds']} --symbolic-sv-size {presets['nanopore-r10']['symbolic_sv_size']} --sd {presets['nanopore-r10']['sd']} --compression wb3",
               default="pe", type=click.Choice(["pe", "pacbio-sequel2", "pacbio-revio", "nanopore-r9", "nanopore-r10", "pacbio", "nanopore"]), show_default=True)
-
 @click.option('--pl', help=f"Type of input reads  [default: {defaults['pl']}]",
               type=click.Choice(["pe", "pacbio", "nanopore"]), callback=add_option_set)
 @click.option('--clip-length', help="Minimum soft-clip length, >= threshold are kept. Set to -1 to ignore [default: {deafults['clip_length']}]", type=int, callback=add_option_set)
@@ -376,7 +379,6 @@ def get_reads(ctx, **kwargs):
                              f"nanopore-r9: --mq {presets['nanopore-r9']['mq']} --paired False --min-support '{presets['nanopore-r9']['min_support']}' --max-cov {presets['nanopore-r9']['max_cov']} --dist-norm {presets['nanopore-r9']['dist_norm']} --trust-ins-len False --symbolic-sv-size {presets['nanopore-r9']['symbolic_sv_size']} --sd {presets['nanopore-r9']['sd']} --divergence {presets['nanopore-r9']['divergence']}."
                              f"nanopore-r10: --mq {presets['nanopore-r10']['mq']} --paired False --min-support '{presets['nanopore-r10']['min_support']}' --max-cov {presets['nanopore-r10']['max_cov']} --dist-norm {presets['nanopore-r10']['dist_norm']} --trust-ins-len False --thresholds {presets['nanopore-r10']['thresholds']} --symbolic-sv-size {presets['nanopore-r10']['symbolic_sv_size']} --sd {presets['nanopore-r10']['sd']}",
               default="pe", type=click.Choice(["pe", "pacbio-sequel2", "pacbio-revio", "nanopore-r9", "nanopore-r10", "pacbio", "nanopore"]), show_default=True)
-
 @click.option('--pl', help=f"Type of input reads  [default: {defaults['pl']}]",
               type=click.Choice(["pe", "pacbio", "nanopore"]), callback=add_option_set)
 @click.option('--clip-length', help="Minimum soft-clip length, >= threshold are kept. Set to -1 to ignore [default: {deafults['clip_length']}]", type=int, callback=add_option_set)

dysgu/merge_svs.pyx

@@ -159,6 +159,8 @@ def consistent_alignment_and_cigars(ei, ej, l_ratio):
 
 
 def jaccard_similarity(set1, set2):
+    if not set1 or not set2:
+        return 0
     intersection = len(set1.intersection(set2))
     union = len(set1.union(set2))
     return intersection / union if union != 0 else 0
@@ -648,7 +650,8 @@ def merge_events(potential, max_dist, tree, paired_end=False, try_rev=False, pic
             best_var_seq = w0.variant_seq
             for k in range(1, len(best)):
                 item = best[k]
-                w0.qnames |= item.qnames
+                if w0.qnames is not None and item.qnames is not None:
+                    w0.qnames |= item.qnames
                 w0.pe += item.pe
                 w0.supp += item.supp
                 w0.sc += item.sc