v0.2.0: Avoid putting fully specified indels into structural variant …

…comparison, which can lead to masked smaller calls on heterozygous deletions. Thanks to Severine Catreux

HISTORY.md

@@ -1,3 +1,9 @@
+## 0.2.0 (17 December 2014)
+
+- Avoid putting fully specified indels into structural variant comparison, which
+  can lead to masked smaller calls on heterozygous deletions. Now do exact
+  comparisons via standard methods. Thanks to Severine Catreux.
+
 ## 0.1.9 (6 November 2014)
 
 - Handle file inputs with identical final file names (but in different

README.md

@@ -160,7 +160,7 @@ provide example starting points and details on available options are below:
                       [full,quick] (default:full)
        params: # Processing parameters associated with this experiment
          max-indel: Maximum indel size to include in non-structural variant
-                    comparisons (default: 30)
+                    comparisons (default: 5000)
          multiple-thresh: Threshold for percentage of comparisons to match
                           to consider two multiple sample variants the same.
                           (default: 1.0)

project.clj

@@ -1,4 +1,4 @@
-(defproject bcbio.variation "0.1.9"
+(defproject bcbio.variation "0.2.0"
   :description "Toolkit to analyze genomic variation data, built on the GATK with Clojure"
   :license {:name "MIT" :url "http://www.opensource.org/licenses/mit-license.html"}
   :dependencies [[org.clojure/clojure "1.5.1"]

src/bcbio/variation/structural.clj

@@ -18,8 +18,9 @@
 (def ^{:private true
        :doc "Default maximum indel size for exact comparisons.
              Initially based on assessment by Gavin Oliver: http://f1000r.es/MsY1QZ
-             Increased to work with longer default read sizes where we expect exact matches"}
-  max-indel 100)
+             Increased to work with longer default read sizes where we expect exact matches
+             and avoid switching into SV calling."}
+  max-indel 5000)
 
 ;; ## Interval tree lookup