let index
~reference_build
~(run_with : Machine.t) =
let open KEDSL in
let reference_fasta =
Machine.get_reference_genome run_with reference_build
|> Reference_genome.fasta in
let mosaik_tool = Machine.get_tool run_with Machine.Tool.Default.mosaik in
let name =
sprintf "mosaik-build-%s" (Filename.basename reference_fasta#product#path) in
let index_result = sprintf "%s.mosaik.dat" reference_fasta#product#path in
let jump_file_result = sprintf "%s.mosaik-index" reference_fasta#product#path in
let mosaik_tmp_dir = (Filename.dirname reference_fasta#product#path) // "mosaik-tmp" in
workflow_node ~name
(single_file ~host:(Machine.(as_host run_with)) index_result)
~edges:[
on_failure_activate (Remove.file ~run_with jump_file_result);
on_failure_activate (Remove.file ~run_with index_result);
depends_on reference_fasta;
depends_on Machine.Tool.(ensure mosaik_tool);
]
~tags:[Target_tags.aligner]
~make:(Machine.run_big_program run_with ~name
~self_ids:["mosaik"; "index"]
Program.(
Machine.Tool.(init mosaik_tool)
&& shf "mkdir -p %s" mosaik_tmp_dir
&& shf "export MOSAIK_TMP=%s" mosaik_tmp_dir
&& shf "MosaikBuild -fr %s -oa %s"
reference_fasta#product#path
index_result
&& shf "MosaikJump -ia %s -hs 15 -out %s"
index_result
jump_file_result
))