Hi @Lara Ianov, I am looking to work on issue #75 and just joined the hackathon. I wanted to run a quick test of the scnanoseq pipeline and need sample fastq files that is known to work with the pipeline. Is there a location you can point me to get a few fastq files to test the pipeline before I start working on the BLAZE issue

Hi ryan, we've got data that gets pulled by both the

test

and

test_full

profiles. (The

test

profile is just a downsampled version of

test_full

). We also use the 10X dataset available here (https://www.10xgenomics.com/datasets/3k-human-squamous-cell-lung-carcinoma-dtcs-chromium-x-2-standard) for validation as well (its much larger fyi)

@Austyn Trull beat me to it but yes! If you run through those files (especially the full ones from the 10X website), you can compare it to what we published in the scnanoseq paper in terms of barcodes etc. these are the datasets that we are at the moment using as a starting points for tests.

okay thank you, I will do this!

The 10X data Austyn linked above is for 5' chemistry, Here is the 3' dataset we use since the 3' vs 5' is a point of relevance to BLAZE https://www.10xgenomics.com/datasets/5k-human-pbmcs-3-v3-1-chromium-controller-3-1-standard

I have a few questions to orientate myself just to make sure I understand the issue. I was able to run the pipeline using test and docker profiles and within the work directory generated from the process correct_barcodes I see this file: less work/8b/95ee65ce6366c0d51b123e3655cd7d/ERR9958134.corrected_bc_umi.tsv From my understanding, you are looking to see how well BLAZE's new barcode correction does when compared to your custom python script (correct_barcodes.py), that does this? If I am thinking about this correctly, I should ensure the blaze process outputs some form of correct file and compare it to the one produced by the correct_barcodes.py? Is this file (putative_bc.csv) generated from the blaze tool what I will be using to compare to the custom file generated (ERR9958134.corrected_bc_umi.tsv)? If this is the case, what is the best way to evaluate the accuracy of barcodes from both methods?

So part of the PR is to understand any outputs that BLAZE now produces. It did not used to provide a correction functionality, so our assumption is that it may now have an output specifically for corrected barcodes. This corrected barcode file is what we would be comparing against our corrected ones to evaluate if we can possibility deprecate our custom method or at least add BLAZE's new functionality as an option until we deprecate the custom process later on. With that being said, how you go about this is certainly up to discussion here. One way, would be to actually just go ahead and make changes to add BLAZE correction to the pipeline and maybe we can validate the outputs later as part of the workflow. If you add this functionality as a PR, Austyn and I can then review the BLAZE implementation as well as the data. So what do you think? Is that clear? (TR;DR: you can 1. simply work towards adding the BLAZE barcode correction into the workflow as a PR and or 2. Validate BLAZE's outputs which again we are happy to do later on if the functionality is present in scnanoseq)

@Austyn Trull I don't see any tests in the repo (at least on the dev branch). Am I missing something?

Hi @Gary Burnett, just for clarification, are you referring to nf-test? We have test-profiles for the moment, nf-test is currently in the template update PR that we have slated to complete for this hackathon

Ahh yeah I was looking for nf-test. I'm implementing my own nf-test for the

align_longreads

subworkflow so I can test my GPU additions to it. But I don't want to duplicate work you've already done.

Got it, it should be good, ill merge your work with the nf-test PR!

Thank you