Hey everyone, I've been running a custom version o...
# methylseq
d
Hey everyone, I've been running a custom version of the methylseq pipeline (originally based on methylseq-3.0.0) for a little while that adds a couple of features to it (Bismark path only): 1. UMI handling (if run in umi mode, it uses fastp to extract UMIs and then umi tools dedup to perform umi-aware deduplication). 2. Adds Picard CollectWGSMetrics if run in wgs mode and Picard CollectHSMetrics otherwise. 3. Ability to run with positive/negative control sequences. If run with --run_controls set and a fasta with positive and negative control sequences added, it will split the aligned reads into three bams, run picard metrics and methylation extractor on each of them, and then generate multiple multiqc reports for bams with positive, negative, and no control sequences. If there is interest in this, I was thinking of contributing these additions back to nf-core. I recently ported my updates from methylseq-3.0.0 to the methylseq-4.2.0-dev branch so it would be compatible. The vast majority of the changes are in the fastq_align_dedup_bismark subworkflow. I know that contributing large changes like this isn't the preferred way to do things, so I was hoping to get some advice from the community on the best way to contribute these changes. I'm also not sure the best way to go about getting additional test data for the workflow (which would ideally be a small dataset with UMIs and/or control sequence data) and to upload a reference with positive/negative control sequences appended. The branch I have for this is here on github: github.com/dmarron/nf-core-methylseq/tree/add-umi-handling-and-controls
a
I mentioned these as a potential feature branch some months ago. We have used it on data locally, but I don't think we have sharable UMI data that we could supply for public use.
b
We were thinking along very similar lines, but haven't made any move yet to integrate this. We are now routinely using UMIs for cfDNA TAPS data, so this has some level of priority for us, along with upgrading the rastair version to 2.x and providing SNP and Indel calls at least in the TAPS path of the pipeline...
Alas, there's a shortage of people on our end to really take this on 🙈