New Features

Some basics:

~/flowr/runs/
- scripts/: place where scripts would be put (if any....)
- runs/: place where all the execution logs etc would be located

load_conf()

Loads a set of parameters into the R environment.

• needs a tab separate file
• if a parameter name ends with a special extension (_exe, _dir, _path, _jar), it would check and give a soft warning
• examples:
  • java_exe
  • gatk_jar
  • picard_dir

search_conf(x="ngsflows.conf")

For a given parameter file name, it searches for all of them and returns them in a sequence they should be executed.

• files this caters to:
  • flowr.conf: basic flowr configuration, like default queue, flow_base_path: not implemented
  • ngsflows.conf: all the basic reflib path shared by different
• sequence: the later overrides former if parameter name matches
  • rpackage: conf folder
  • ~/flowr/conf/my.conf
  • a full path to a con file, overrides all of the above

creating flow_mat

• structure, keep it clean
  • aln_merge_func.R: all the functions
  • aln_merge_def.txt: flow_definition file
  • A function by the same name (aln_merge), accepts arguments and creates a data_frame flow_mat

Submission

• One can save this to file, or directly supplied to to_flow() along with the definition.

run(flowname, ...)

runs a specific (work) flow, searched using the supplied flowname.

Once the flow_mat creation is stable, one can take this to production level. Here run() is used to automatically find the flowname_func.R, and also use flowname_def.txt as its accompanying flow_definition

• For example if flowname=aln_merge
• Check if this function already exists in the package ngsflows.
• pipelines may sit in ngsflows/inst/examples
• def: One can always specify a custom def_file and this overrides the default

do_align(execute):

• check config for file exists
• runs/test_bwa/flow_cmds.txt
• put flow_def here.
• Alternatively:
• submit_flow(to_flow(x, def))

important changes to submit_flow

• Add job names to stdout
• DONE !