Meeting notes from today! - Evangelos and James w...
# meta-omics
j
Meeting notes from today! • Evangelos and James will do introductory teaching at the Hackathon (Athens and Berlin) • Chaining: Use output directly from channels rather than output from processes. See publish output in Nextflow (https://www.nextflow.io/docs/latest/workflow.html#publishing-outputs). (Edge release of Nextflow possibly needed) • Samuel mentioned he's looking at ideas for provenance tracking of files. Still WIP/research project but possibly connected to
nf-prov
. • James is working on making a smaller taxon dump • Jennifer applied Bakta on lots of isolates of a bacterial species (want to find genetic features causing pathogenicity in cohort), but gene names can't be used because they get heterogeniously updated in each strain (gene name is based on each annotation tool) - does anyone have any experience with this problem when trying to do gene enrichment. UniProt IDs have been recommended, but this has their own issues. ◦ Evangelos suggested trying: https://biit.cs.ut.ee/gprofiler/orth or https://pavlopoulos-lab-services.org/shiny/app/flame ◦ Daniel remembers a poster on an nf-core pipeline using OrthoFinder ◦ Samuel: maybe ENSEMBL IDs might be better (as they are genes rather than proteins) -> https://ensembl.org/info/genome/stable_ids/index.html • Pipeline updates: ◦ mag: release today ◦ createtaxdb: progressing, battle testing with Sam Wilkinson ◦ taxprofiler: sofia did template merge, sofia also working on sylph, open PR improving resume by sorting krakenuniq inputs prior batching ◦ funcscan: release tomorrow 🎉 ◦ proteinfamilies: first release a couple of weeks ago! Working on minor fixes. James forgot to Zenodo update 👀, working on workflow benchmarking figure ◦ metatdenovo: close to first release! ◦ magmap: template updates and more for 1.1 release, PR out now: https://github.com/nf-core/metatdenovo/pull/334, needs one more reviewer ◦ phyloplace: 2.0 release PR soon, major rework • PR reviewing trade ◦ Jasmin ◦ Daniel ◦ Danilo
n
Jennifer applied Bakta on lots of isolates of a bacterial species (want to find genetic features causing pathogenicity in cohort), but gene names can't be used because they get heterogeniously updated in each strain (gene name is based on each annotation tool) - does anyone have any experience with this problem when trying to do gene enrichment. UniProt IDs have been recommended, but this has their own issues.
I would also recommend to cluster genes using orthofinder. People have also used k-mer based approaches, this paper uses both https://www.nature.com/articles/s41467-024-49643-w
It is not about pathogenicity per se, but it is still about picking up genetic basis of a trait
In the simplest case, you may be lucky with computing all-vs-all orthogroups and checking for OG presence/absence correlated with your trait of interest
🙏 1
j
@Jennifer Müller 👆