search

HurlesGroupSanger / indelible

Structural Variation breakpoint discovery via adaptive learning
GNU General Public License v3.0
15 stars 1 forks source link

hg38 data or instructions on generation of data resources #1

Closed ameynert closed 3 years ago

ameynert commented 4 years ago

Could the developers provide a hg38 data zip file or instructions on how to generate a custom data folder. If the latter, I will generate for hg38 myself and share.

eugenegardner commented 4 years ago

Hello,

I am working on getting an Hg38 data package together and will let you know when it has been pushed live.

eugenegardner commented 4 years ago

Hello,

I have generated a pull request (#2) for code/data which should get Hg38 working with InDelible. I have:

  1. Added a hg38 data zip into the ./data/ folder. This does not include the reference genome itself, or the wmasker.obinary file necessary for the blast step.
  2. Modified the README to (hopefully) provide good instructions on getting InDelible to work with Hg38.
    • This now includes a section in "Output" about how to perform recommended default filtering as we have done in DDD
  3. Modified the code base to be more adaptable to non-hg19 references.
  4. Made --config a required option for ALL InDelible commands, which helps to ease the transition to Hg38.
  5. Tested InDelible on both an Hg19 and Hg38 Decipher Developmental Disorders .cram file, which generated expected output (single de novo variant deletion within MECP2).

Please let me know if you are successful in getting InDelible to run with your data. Make sure to perform a new git checkout of InDelible and switch to the "hg38" branch.

ameynert commented 3 years ago

It's working fine, thank you for the detailed instructions @eugenegardner. I've written a Nextflow pipeline to run the steps as vr-runner was not immediately installable on the system I'm using. Currently the nextflow.config file is set to use PBSpro and has one hard-coded path, also the R scripts for filtering assume Rscript is at /usr/bin/Rscript, so it's not 100% portable yet. https://github.com/ameynert/nf-indelible