tool wishlist

[a-frantz]

This issue is meant for tracking tools we'd like to see WDL implementations for (but that we aren't actively working on adding). For any intrepid contributors, adding one of these tools to the repo could be a good first issue :)

• Salmon (https://salmon.readthedocs.io/en/latest/salmon.html)
• squab (https://github.com/zaeleus/squab)
• fastp (https://github.com/OpenGene/fastp) feat: support fastp #244

This list will be updated periodically as we think of other neat tools that should have WDL implementations. Feel free to comment here with anything you'd like to see added!

[PriyankaaXD]

Hi! I'd like to try adding a WDL implementation for Salmon from this list. This would be my first contribution to this project. I'm going to start by reading through an existing tool task in the repo to understand the pattern you use, then write a draft. If it's helpful, I can share my plan here before I write the full implementation, in case I'm heading in the wrong direction.

[a-frantz]

Hi! I'd like to try adding a WDL implementation for Salmon from this list. This would be my first contribution to this project. I'm going to start by reading through an existing tool task in the repo to understand the pattern you use, then write a draft. If it's helpful, I can share my plan here before I write the full implementation, in case I'm heading in the wrong direction.

Sounds great! We would appreciate it!

[PriyankaaXD]

Update on my approach: I'm planning two WDL tasks in tools/salmon.wdl — build_salmon_index (wraps salmon index) and quant (wraps salmon quant), following the same two-task pattern used in tools/star.wdl (a build/index task + a run task). For the library type, I'm defaulting -l to A (auto-detect) to keep it simple for most users. Planning to use an existing BioContainers Salmon image. Does this overall approach sound right, or is there anything you'd steer differently before I finalize and test it?

[a-frantz]

That all sounds correct to me! Although I have some more detailed guidance after briefly familiarizing myself with the Salmon docs:

• the "important options" called out in the docs should also be exposed as inputs in addition to library type.
  • All of these options should have WDL default values matching the tool's default value.
  • For documentation, copy and pasting from the docs into the meta sections is adequate; no need to editorialize or rephrase the docs which have already been written!
• salmon quant has both a "mapping mode" and an "alignment mode"
  • these take different input formats (mapping->fast{q,a} or alignment->bam), and right now we only care to implement the "mapping mode". You can ignore anything in salmon which is exclusive to alignment mode; the salmon_quant WDL task should take one or two FASTQs as input (and there should not be any BAM input accepted)

don't hesitate to ping me on this thread if you run into any issues while writing the WDL wrapper :)

[PriyankaaXD]

I saw the repo recently started migrating from pytest-workflow to Sprocket's own test tooling (the "drop pytest in favor of Sprocket test" PR), but it looks like that's still WIP and not finished. Should I write a pytest-workflow test per the current CONTRIBUTING.md, hold off on tests until the new format solidifies, or is there a specific format you'd like me to try? Happy to do whatever's most useful.

[a-frantz]

@PriyankaaXD sorry about that! The README was out of date but we just fixed it - #323

Please implement tests using sprocket dev test! https://sprocket.bio/subcommands/test.html

disclaimer, I'm the primary designer of the Sprocket test framework, so if you have any feedback on the test writing/running process, don't hesitate to share :)

[PriyankaaXD]

Opened a PR for this: [https://github.com//pull/326]

Implements build_salmon_index and quant tasks for Salmon in mapping-mode only, with tests using the new Sprocket test framework. Would appreciate a review whenever you get a chance — happy to make any changes needed!