0.4.8

• Now double-nested arrays are supported for input files for both CWL and WDL.
• Fields related to WDL in the the input json (job description json) has been changed slightly:
  • wdl_filename -> wdl_main_filename
  • New field wdl_child_filenames is required now for WDL (can be [] if there are no child WDL files)

0.4.7

• Nested file array input support (e.g. [[f1, f2], [f3, f4]]) for both CWL and WDL.

0.4.6

Tibanna unicorn now supports Simple WDL

Constraints

• Input files must be either a singlet or an array (e.g. not an array of arrays or a dictionary, etc).
• All outputs must be files (no streaming, string, integer etc allowed).
• If an input file is used by multiple steps, it will be downloaded multiple times to Tibanna's AWSEM EC2 instance.
• Secondary files are not supported. (WDL itself does not support secondary files)
• Compared to CWL, for WDL, Tibanna does not yet parse commands into postrun.json and it also does not calculate md5sum for output files.

0.4.5

• New command invoke kill available to kill a specific job (through execution arn)
• Both invoke kill and invoke kill_all now kill both step function run and EC2 instances.

0.4.4

• Tibanna Pony (4dn) : fastqc trigger as a dependent of md5 trigger (fastqc will start run_task once md5 finishes successfully)

0.4.3

• Response from function run_workflow now contains jobid.

from core.utils import run_workflow
res = run_workflow(some_job)
jobid = res['jobid']

• Cli invoke run_workflow now prints out jobid.

0.4.1

• Now python setup.py install can also be used to install tibanna as an alternative to pip install -r requirements.txt.
• Some dependency versions have changed. Update existing installation using pip install -r requirements.txt --no-cache-dir.

0.4.0

• The structure of the directory awsf has been changed. Tibanna (specifically the run_task_awsem lambda) must be redeployed to use this changed environment, invoke deploy_core run_task_awsem --suffix=<suffix> must be run ( is corresponds to the part after ''tibanna_unicorn_' ('tibanna_unicorn_')

• 'copy_to_s3' and 'launch_instance' fields are removed from the 'config' field of the input json (they were always true).

• arguments '--workflow' is now replaced by '--sfn' for run_workflow, rerun, rerun_many and kill_all

• Pony:

  • It's compatible with the new system in fourfront where file formats are objects. (e.g. it gets file format-extension mapping from file_format objects instead of file_fastq, file_reference and file_processed schemas.)

0.3.9

• Profile-based input data bucket permission is added for unicorn.
  • TIBANA_PROFILE_ACCESS_KEY and TIBANNA_PROFILE_SECRET_KEY must be set and exported at deployment.
  • For individual input file, add profile field ("profile": "user1") in addition to bucket_name and object_key.
• For pony, custom wfr field can be added using wfr_meta field.

0.3.8

• Now one can set job dependency between multiple tibanna runs using the 'dependency' field in the input json.