v0.1.5: Finalize variant evaluation support with new bcbio-nextgen ch…

…anges: handle bgzip inputs, support non-GATK callable regions, avoid VariantEval due to java errors. Thanks to Severine Catreux.

HISTORY.md

@@ -1,8 +1,12 @@
-## 0.1.5 (In progress)
+## 0.1.5 (15 March 2014)
 
 - Move to MIT licensed GATK 3.0 framework.
+- Support bgzipped inputs for variant assessment. Thanks to Severine Catreux.
 - Support lightweight loading options for gemini integration to avoid large load
   times with new gene tables in gemini 0.6.5.
+- Avoid running GATK VariantEval which causes intermittent java core dumps.
+- Avoid re-runs of callable regions when already prepared using non-GATK chanjo-based
+  methods.
 
 ## 0.1.4 (5 March 2014)
 

README.md

@@ -33,7 +33,7 @@ associated with different variant representations.
 
 ### Download
 
-The latest release is 0.1.4 (5 March 2014): [bcbio.variation-0.1.4-standalone.jar][dl].
+The latest release is 0.1.5 (15 March 2014): [bcbio.variation-0.1.5-standalone.jar][dl].
 Run from the command line:
 
     $ java -jar bcbio.variation-VERSION-standalone.jar [arguments]
@@ -44,7 +44,7 @@ the library for variant comparison, normalization and ensemble calling.  Note
 that bcbio.variation requires Java 1.7 since the underlying GATK libraries are
 not compatible with earlier versions.
 
-[dl]: https://github.com/chapmanb/bcbio.variation/releases/download/v0.1.4/bcbio.variation-0.1.4-standalone.jar
+[dl]: https://github.com/chapmanb/bcbio.variation/releases/download/v0.1.5/bcbio.variation-0.1.5-standalone.jar
 
 ### As a library
 

project.clj

@@ -1,4 +1,4 @@
-(defproject bcbio.variation "0.1.5-SNAPSHOT"
+(defproject bcbio.variation "0.1.5"
   :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/callable.clj

@@ -31,7 +31,8 @@
     (if-not (fs/exists? base-dir)
       (fs/mkdirs base-dir))
     (broad/index-bam align-bam)
-    (broad/run-gatk "CallableLoci" args file-info {:out [:out-bed :out-summary]})
+    (when (itx/needs-run? (:out-bed file-info))
+      (broad/run-gatk "CallableLoci" args file-info {:out [:out-bed :out-summary]}))
     (:out-bed file-info)))
 
 (defn features-in-region [source space start end]

src/bcbio/variation/compare.clj

@@ -157,18 +157,20 @@
     (let [c-files (select-by-concordance (:sample exp) c1 c2 (:ref exp)
                                          :out-dir (get-in config [:dir :out])
                                          :intervals (:intervals exp))
-          eval (calc-variant-eval-metrics (:file c1) (:file c2) (:ref exp)
-                                          :out-base (first c-files)
-                                          :intervals (:intervals exp))
-          c-eval (calc-variant-eval-metrics (:file c1) (:file c2) (:ref exp)
-                                            :out-base (fsp/add-file-part (first c-files) "callable")
-                                            :intervals (callable-intervals exp c1 c2))]
+          ;; eval (calc-variant-eval-metrics (:file c1) (:file c2) (:ref exp)
+          ;;                                 :out-base (first c-files)
+          ;;                                 :intervals (:intervals exp))
+          ;; c-eval (calc-variant-eval-metrics (:file c1) (:file c2) (:ref exp)
+          ;;                                   :out-base (fsp/add-file-part (first c-files) "callable")
+          ;;                                   :intervals (callable-intervals exp c1 c2))
+          ]
       {:c-files (zipmap-ordered (map keyword
                                      ["concordant" (discordant-name c1) (discordant-name c2)])
                                 c-files)
        :c1 c1 :c2 c2 :exp exp :dir (config :dir)
-       :metrics (report/concordance-report-metrics (:sample exp) eval)
-       :callable-metrics (report/concordance-report-metrics (:sample exp) c-eval)})))
+       ;; :metrics (report/concordance-report-metrics (:sample exp) eval)
+       ;; :callable-metrics (report/concordance-report-metrics (:sample exp) c-eval)
+       })))
 
 (defn compare-two-vcf
   "Compare two VCF files, handling standard and haploid specific comparisons."

src/bcbio/variation/normalize.clj

@@ -551,6 +551,14 @@
      (neg-qual? xs) []
      :else xs)))
 
+(defn zipsafe-reader
+  "Provide a reader, handling gzipped or plain text inputs."
+  [f]
+  (reader
+   (cond
+    (.endsWith f ".gz") (-> f input-stream java.util.zip.GZIPInputStream.)
+    :else f)))
+
 (defn clean-problem-vcf
   "Clean VCF file which GATK parsers cannot handle due to illegal characters.
   Fixes:
@@ -559,7 +567,7 @@
     - Removes spaces in INFO fields."
   [in-vcf-file ref-file sample call & {:keys [out-dir]}]
   (let [get-ref-base (ref-base-getter ref-file)
-        out-file (fsp/add-file-part in-vcf-file "preclean" out-dir)]
+        out-file (string/replace (fsp/add-file-part in-vcf-file "preclean" out-dir) ".vcf.gz" ".vcf")]
     (letfn [(remove-gap [n xs]
               (assoc xs n
                      (-> (nth xs n)
@@ -584,7 +592,7 @@
                      (string/join "\t"))))]
       (when (itx/needs-run? out-file)
         (itx/with-tx-file [tx-out-file out-file]
-          (with-open [rdr (reader in-vcf-file)
+          (with-open [rdr (zipsafe-reader in-vcf-file)
                       wtr (writer tx-out-file)]
             (doall
              (map #(.write wtr (str % "\n"))