Skip to content

Latest commit

 

History

History
90 lines (53 loc) · 3.73 KB

CHANGELOG.md

File metadata and controls

90 lines (53 loc) · 3.73 KB

sheynkman-lab/Long-Read-Proteogenomics: Changelog

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

NOTE: For an example on where to start counting from in dev phases of development, consult this informative link

v0.1.0

PR: #103

Initial release of sheynkman-lab/Long-Read-Proteogenomics, created with the nf-core template.

Added

  • nf-core based repository structure
  • modules and additions to the individual scripts

v0.2.0

PR: #132

Enables use of --config as a parameter

Added

  • Enables use of --config as a parameter
  • Enables use of compressed .fa.gz gencode_translation_fasta in the process make_gencode_database

Fixed

  • Fixes test config, add https ZENODO links as env agnostic inputs (runs in cloud, local, cluster, gh-actions)
  • Updates initial channel definitions to allow for informative error messages

Dependencies

None added

(Optional new version)

PR: #140

Added documentation, scripts to pull data from AWS buckets to Zenodo.

The main reasons why ZENODO vs AWS S3 is a good idea:

  1. Data versioning (number 1 important reason). In S3 data can be overwritten for the same path at any point possibly breaking the pipeline.
  2. Cost, removing data from S3. These data are tiny but the principle stays: The less storage the better
  3. Access, Academic users familiarity/accessibility - Most reviewers, readers of the pipeline and paper will know ZENODO and will be able to use the data, AWS has an entry barrier for this group of people

Added

  • AWStoZenodo.md - detail steps for downloading data files from Amazon S3 buckets and upload procedure to Zenodo
  • download_aws.sh - documentation of the files that were extracted from Amazon S3 buckets.
  • upload_to_zenodo.sh - documentation of the files that were deposited into Zenodorepository for use by theLong-Read-Protegenomics` pipeline.
  • download_and_reconstruct.sh - script to download files from Zenodo for use in execution of the pipeline.

Fixed

None: To Do: Modify cloudos_jurkat_merged_bam.config file to use the files from Zenodo in the execution of the main.nf pipeline

Dependencies

v0.2.1

PR: #145

Added

  • Optional input into pipeline - metamorpheus_toml for user controlled toml file used in search

Fixed

  • novel peptide: set had 'append' rather than 'add' command - bug from change from list to set after incorporating UniProt

Dependencies

None Added

Dependencies

v0.2.2

PR: #153

Added

  • output from the full test with running sqanti and running with the two fastq files now on Zenodo with a new version, DOI. A smaller version of star limited to only chr22 was added to the test repository - all tests run from laptop as well as with GitHub actions.

Fixed

-- all conf files now updated with the new latest version number, test.config (which is redundant with test_with_sqanti.config), test_with_sqanti.config and test_without_sqanti.config'

Dependencies

None added