wip

src/cljam/io/cram/decode/data_series.clj

@@ -115,6 +115,7 @@
 (defn build-tag-decoders [{:keys [tags]} blocks]
   (let [content-id->block-data (into {} (map (juxt :content-id :data)) blocks)]
     (reduce-kv (fn [decoders tag {tag-type :type :keys [encoding]}]
-                 (let [decoder (build-tag-decoder encoding tag-type content-id->block-data)]
-                   (assoc decoders tag (fn [] {:type tag-type :value (decoder)}))))
+                 (let [decoder (build-tag-decoder encoding tag-type content-id->block-data)
+                       tag-type' (str (if (#{\c \C \s \S \i \I} tag-type) \i tag-type))]
+                   (assoc decoders tag (fn [] {:type tag-type' :value (decoder)}))))
                {} tags)))

src/cljam/io/cram/decode/record.clj

@@ -38,10 +38,11 @@
       (let [mate-flag (long (MF))
             bam-flag (cond-> (long (:flag record))
                        (pos? (bit-and mate-flag 0x01)) (bit-or 0x20)
-                       (pos? (bit-and mate-flag 0x02)) (bit-or 0x08))]
+                       (pos? (bit-and mate-flag 0x02)) (bit-or 0x08))
+            rnext (get-in cram-header [:SQ (NS) :SN])]
         (assoc record
                :flag bam-flag
-               :rnext (get-in cram-header [:SQ (NS) :SN])
+               :rnext (if (= (:rname record) rnext) "=" rnext)
                :pnext (NP)
                :tlen (TS)))
       (cond-> record
@@ -61,7 +62,7 @@
              (assoc record :options))))))
 
 (defn- record-seq
-  [seq-reader {:keys [preservation-map]} {:keys [rname pos ::end ::len]} features]
+  [seq-reader {:keys [preservation-map]} {:keys [rname pos end ::len]} features]
   (let [region {:chr rname :start pos :end end}
         ref-bases (.getBytes ^String (cseq/read-sequence seq-reader region))
         bs (byte-array len (byte (int \N)))
@@ -182,7 +183,7 @@
       (if (zero? i)
         (let [features (persistent! fs)
               end (record-end record features)
-              record' (assoc record ::end end)]
+              record' (assoc record :end end)]
           (assoc record'
                  :seq (record-seq seq-reader compression-header record' features)
                  :qual (record-qual record' features decoders)
@@ -213,7 +214,10 @@
     (let [bb (bb/allocate-lsb-byte-buffer len)
           _ (dotimes [_ len]
               (.put bb (byte (BA))))
-          record' (assoc record :seq (String. (.array bb)))]
+          record' (assoc record
+                         :seq (String. (.array bb))
+                         :mapq 0
+                         :cigar "")]
       (if (zero? (bit-and (long flag) 0x01))
         record'
         (let [bb (bb/allocate-lsb-byte-buffer len)]

src/cljam/io/cram/reader.clj

@@ -16,12 +16,13 @@
 
                  (pos? (bit-and mate-flag 0x08))
                  (bit-or 0x08))
-         :rnext (:rname mate)
+         :rnext (if (= (:rname record) (:rname mate))
+                  "="
+                  (:rname mate))
          :pnext (:pos mate)))
 
 (defn- update-mate-records
-  [{^long s1 :pos ^long e1 ::record/end :as r1}
-   {^long s2 :pos ^long e2 ::record/end :as r2}]
+  [{^long s1 :pos ^long e1 :end :as r1} {^long s2 :pos ^long e2 :end :as r2}]
   (let [r1' (update-next-mate r1 r2)
         r2' (update-next-mate r2 r1)]
     (if (or (pos? (bit-and (long (:flag r1)) 0x04))
@@ -63,7 +64,7 @@
     (dotimes [i n]
       (aset records i (record-decoder)))
     (resolve-mate-records records)
-    (map #(dissoc % ::record/flag ::record/len ::record/end ::record/next-fragment)
+    (map #(dissoc % ::record/flag ::record/len ::record/next-fragment)
          records)))
 
 (defn- read-container-records [^Buffer bb seq-reader cram-header container-header]