Biokepi API : Biokepi_pipeline_edsl.To

let concat : Annotated_file.t -> Annotated_file.t =
    fun l ->
      begin match get_list (AF.get_file l) with
      | Fastq one :: [] ->
        Fastq one
        |> AF.with_provenance "concat" ["single-element", AF.get_provenance l]
      | Fastq first_fastq :: _ as lfq ->
        let fqs = List.map lfq ~f:get_fastq in
        let r1s = List.map fqs ~f:(KEDSL.read_1_file_node) in
        let r2s = List.filter_map fqs ~f:KEDSL.read_2_file_node in
        (* TODO add some verifications that they have the same number of files?
           i.e. that we are not mixing SE and PE fastqs
        *)
        let concat_files ~read l =
          let result_path =
            Name_file.in_directory
              Config.work_dir
              ~readable_suffix:(
                sprintf "%s-Read%d-Concat.fastq"
                  first_fastq#product#escaped_sample_name read) (
              first_fastq#product#escaped_sample_name
              :: first_fastq#product#fragment_id_forced
              :: List.map l ~f:(fun wf -> wf#product#path)
            )
          in
          Workflow_utilities.Cat.concat ~run_with l ~result_path in
        let read_1 = concat_files r1s ~read:1 in
        let read_2 =
          match r2s with [] -> None | more -> Some (concat_files more ~read:2)
        in
        Fastq (
          KEDSL.fastq_node_of_single_file_nodes ~host
            ~name:first_fastq#product#sample_name
            ~fragment_id:"edsl-concat"
            read_1 read_2
        )
        |> AF.with_provenance "concat" ["fastq-list", AF.get_provenance l]
      | other ->
        ksprintf failwith "To_workflow.concat: not implemented"
      end