feat: allow setting of custom alignment properties

.test/config-chm-eval/config.yaml

@@ -184,3 +184,8 @@ report:
   stratify:
     activate: false
     by-column: condition
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

.test/config-giab/config.yaml

@@ -157,3 +157,8 @@ params:
 
 gene_coverage:
   min_avg_coverage: 5
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

.test/config-no-candidate-filtering/config.yaml

@@ -123,3 +123,8 @@ report:
   stratify:
     activate: false
     by-column: condition
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

.test/config-simple/config.yaml

@@ -127,3 +127,8 @@ tables:
     coverage: true
     event_prob: true
   generate_excel: true
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

.test/config-sra/config.yaml

@@ -112,4 +112,9 @@ tables:
     genotype: true
     coverage: true
     event_prob: true
-  generate_excel: true
+  generate_excel: true
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

.test/config-target-regions/config.yaml

@@ -126,3 +126,8 @@ tables:
     coverage: true
     event_prob: true
   generate_excel: true
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

.test/config-target-regions/config_multiple_beds.yaml

@@ -124,3 +124,8 @@ tables:
     coverage: true
     event_prob: true
   generate_excel: true
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

.test/config_primers/config.yaml

@@ -115,4 +115,9 @@ tables:
     genotype: true
     coverage: true
     event_prob: true
-  generate_excel: true
+  generate_excel: true
+
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: ""

config/alignment_properties.tsv

@@ -0,0 +1 @@
+name	path

config/config.yaml

@@ -284,3 +284,12 @@ params:
   freebayes:
     min_alternate_fraction: 0.05 # Reduce for calling variants with lower VAFs
     extra: ""
+
+# If activated preprocessed alignment properties can be applied to each sample individually.
+# Paths to the alignment properties json-files need to be set in a tsv-file containing a property name and path.
+# Alignment properties will be derived from a customizable column in the sample sheet.
+# If not property name is set for a sample the alignment properties will be estimated best on the samples mapping.
+custom_alignment_properties:
+  activate: false
+  column: "panel"
+  tsv: "config/alignment_properties.tsv"

workflow/rules/calling.smk

@@ -47,7 +47,7 @@ rule varlociraptor_preprocess:
         candidates=get_candidate_calls,
         bam="results/recal/{sample}.bam",
         bai="results/recal/{sample}.bai",
-        alignment_props="results/alignment-properties/{group}/{sample}.json",
+        alignment_props=get_alignment_props,
     output:
         "results/observations/{group}/{sample}.{caller}.{scatteritem}.bcf",
     params:

workflow/rules/common.smk

@@ -112,6 +112,29 @@ primer_panels = (
 )
 
 
+def is_activated(xpath):
+    c = config
+    for entry in xpath.split("/"):
+        c = c.get(entry, {})
+    return bool(c.get("activate", False))
+
+
+custom_alignment_props = (
+    (
+        pd.read_csv(
+            config["custom_alignment_properties"]["tsv"],
+            sep="\t",
+            dtype={"name": str, "path": str},
+            comment="#",
+        )
+        .set_index(["name"], drop=False)
+        .sort_index()
+    )
+    if is_activated("custom_alignment_properties")
+    else None
+)
+
+
 def get_calling_events(calling_type):
     events = [
         event
@@ -606,13 +629,6 @@ def get_all_group_observations(wildcards):
     )
 
 
-def is_activated(xpath):
-    c = config
-    for entry in xpath.split("/"):
-        c = c.get(entry, {})
-    return bool(c.get("activate", False))
-
-
 def get_star_read_group(wildcards):
     """Denote sample name and platform in read group."""
     platform = extract_unique_sample_column_value(wildcards.sample, "platform")
@@ -1591,3 +1607,14 @@ def get_delly_excluded_regions():
         )
     else:
         return []
+
+
+def get_alignment_props(wildcards):
+    if is_activated("custom_alignment_properties"):
+        alignment_prop_column = config["custom_alignment_properties"]["column"]
+        prop_name = extract_unique_sample_column_value(
+            wildcards.sample, alignment_prop_column
+        )
+        if pd.notna(prop_name):
+            return custom_alignment_props.loc[prop_name, "path"]
+    return f"results/alignment-properties/{wildcards.group}/{wildcards.sample}.json"

workflow/rules/mapping.smk

@@ -15,8 +15,6 @@ rule map_reads_bwa:
         "v3.8.0/bio/bwa/mem"
 
 
-# Create distance and minimizer index before mapping
-# Otherwise it will be performed on the first execution leading to race conditions for multiple samples
 rule map_reads_vg:
     input:
         reads=get_map_reads_input,
@@ -31,7 +29,7 @@ rule map_reads_vg:
         extra="",
         sorting="fgbio",
         sort_order="queryname",
-    threads: 8
+    threads: 64
     wrapper:
         "v5.3.0/bio/vg/giraffe"
 

workflow/schemas/config.schema.yaml

@@ -347,6 +347,17 @@ properties:
       - freebayes
       - varlociraptor
 
+  custom_alignment_properties:
+    type: object
+    properties:
+      activate:
+        type: boolean
+      column:
+        type: string
+      tsv:
+        type: string
+
+
 required:
   - samples
   - units
@@ -355,3 +366,4 @@ required:
   - calling
   - params
   - annotations
+  - custom_alignment_properties