update on sto reading

gistfile1.clj

(defn join-sto-fasta-lines [infilespec origin]
  (let [[seq-lines gc-lines] (sto-GC-and-seq-lines infilespec)
        
        gc-lines (if (not= origin "")
                   (concat (take 1 gc-lines) [origin] (drop 1 gc-lines))
                   gc-lines)
        recombined-seqs (sort-by
                         #(-> % second first)
                         (vec
                          (reduce
                           (fn [m l]
                             (let [[nm sq] (if (.startsWith l "#")
                                             (str/split #"\s{2,}+" l)
                                             (str/split #"\s+" l))
                                   prev (get m nm [(gen-uid) ""])]
                               (assoc m  nm [(first prev)
                                             (str (second prev) sq)])))
                           {} seq-lines)))
        {seq-lines false cons-lines true} (group-by
                                           #(or (.startsWith (first %) "//")
                                                (.startsWith (first %) "#"))
                                           recombined-seqs)]
    [gc-lines seq-lines cons-lines]))


(defn join-sto-fasta-file
  "Block/join unblocked sequence lines in a sto or fasta file.  For
   sto files ORIGIN is a #=GF line indicating tool origin of file.
   For example, '#=GF AU Infernal 1.0.2'.  Defaults to nothing."

  [in-filespec out-filespec
   & {origin :origin :or {origin ""}}]
  (let [[gc-lines seq-lines cons-lines] (join-sto-fasta-lines in-filespec origin)]
    (io/with-out-writer (fs/fullpath out-filespec)
      (doseq [gcl gc-lines] (println gcl))
      (doseq [sl seq-lines]
        (let [[nm [id sq]] sl]
          (cl-format true "~A~40T~A~%" nm sq)))
      (doseq [cl cons-lines]
        (let [[nm [id sq]] cl]
          (cl-format true "~A~40T~A~%" nm sq))))))