mag branch 'minigut' is the same sample as modules branch 'bacteroides_fragilis'

[werner291]

Hello, people of nf-core. While working on my project to rewrite MaxBin2, I encountered a bit of an odd issue.

I was benchmarking a gene caller on two test datasets: the B. fragilis data from the modules branch, and the "minigut" data from the mag branch. I got identical results on both and dug into why.

Turns out the reads are the same file — test_data/test_minigut_R1.fastq.gz on mag and data/genomics/prokaryotes/bacteroides_fragilis/illumina/fastq/test1_1.fastq.gz on modules have the same SHA-256 (sha256-qQ9NSKMVSIbnKlIhuN1rEVnqCQ5dviBjTpZbTugBFqs=). The contigs are different files but contain the same 272 sequences in a different order

Looking at the git history, the reads were first committed to mag in 2018 (2c50b6d5, "test gut bact"), then the same file showed up on modules under bacteroides_fragilis/ in 2021 (b2197bc5). The contigs were re-assembled and added to mag in 2023 (95f022cc).

The naming is what tripped me up — "minigut" sounds like a multi-species gut community, not a single-organism B. fragilis dataset. There's nothing on either branch that connects the two. Maybe worth a note in the mag branch README?

[jfy133]

Hi @werner291

Good eyes!

Yes, the original mag dataset was a 'synthetic' dataset derived from a common gut bacteria.

When I took over the project and started porting over local to official nf-core modules and adding tests, I made the same discovery that it was actually a single-taxon 'sample'... thus the rename on the modules repository.

I was hoping at some point the original authors would be able to so some archaeology and give more history behind the minigut samples, but I think that time is past. I'll add a little note.