Steven Robbins

I get in a bit of a huff at the vocal experimentalist crowd that seem to want to discount genomics studies as a blanket cultural norm without individual consideration.

Like, try to tell me a whole class of bacteria all maintain that amount of co-localized machinery without ever using it. It’s possible, but incredibly unlikely. And in fact, SIP studies on DOM degradation were done after this that showed the same taxa we proposed to degrade DOM.

But I agree, everything becomes much more interesting/defensible when you focus on genomes rather than individual genes. At least then I can say “this group is enriched in whole classes of CAZymes, co-localized on the genome in PULs, with transporters for the degrading products, etc.”

Which is a baffling thought process to me. Gene presence and phylogeny are like 99.9% correlated. It’s like, “is the organism selected because of its genes or are the genes enriched because the organism is?” To which I think, what’s the difference?

Actually though, the genes vs organisms thing has been brought up by reviewers of multiple papers in the form of “how do you know those genes aren’t enriched simply because those organisms are enriched? Is it really because the genes are important?”

But I agree. I think at its core, I just think a lot of genomics is treaty superficially and without a starting goal. We try to think of these things like in-silico experiments with expected outcomes based on current theory, rather than “I wonder what gene go up/down?”

Thanks man! I’m actually proud of this one specifically. We got desk rejected 3 times from glam journals with a perfunctory “it’s only genomics, we don’t care” from editors not in the field. Our ISME reviewers in the field had things nice enough to say I wanted to frame the comments.

Finishing a complete giraffe genome from telomere to telomere with Verkko-Fillet High-quality genome assemblies are essential for understanding speciation, evolution, and for building reference genomes and pan-genomes. Despite major advances in long-read sequencing technologies an...

“Finishing a complete giraffe genome from telomere to telomere with Verkko-Fillet” walks you through the process of finishing a complete T2T genome using a new Python toolkit developed by Juhyun Kim that makes Verkko output more accessible... www.biorxiv.org/content/10.1...

Won’t get any complaints from me on that. Lots of people out there with sequence data making stretched claims based on cursory analysis of a few isolated genes.

This was a great read. I feel similarly.

Also throws away a lot of potential to identify new functions in organism with atypical pathways because we’re relying on what’s been seen before in lab rat microbes, but I don’t think pure genomics should typically be the domain of proposing new pathways anyway. Leave that for wet lab experiments.

Then if you go to the KEGG page for each KO, it lists experimental evidence for function.

Still room for error, proteins could still have promiscuous function, but at least if we require that most pathways/module genes are present, with key genes, narrows down to a much more trustworthy set.

That’s actually a great question. It’s why we required the “key” genes, and >70% of the pathway to be present, because many genes belong to multiple pathways. But for example, carbon fixation pathways tend to have key genes that aren’t part of other pathways,so we required those to be present.

Accepted version of our article "Half of microbial eukaryote literature focuses on only twelve human parasites" now out in ISMEJ!

Awesome effort led by undergrad Joanna Lepper and with @hbrappap.bsky.social.

Perfect, thanks Florian! Should have just assumed Anvio would have something. Such a Swiss Army knife these days.

hear*

Unfortunately, the power of EnrichM was the consideration of KOs into modules/pathways and doing the stats on the different DBs with standard outputs. Really wish it got published but Joel got an industry job. Maybe one can hack the backend DBs, but my guess is the formats have changed.

Thanks @jcamthrash.bsky.social and @acritschristoph.bsky.social! DRAM I know of, though it's slowwww from what I here--default to single threads for things that could be multi-threaded. Not sure if that's been fixed. There's also Metacerberus for faster annotation.

I think that's why KEGG is so powerful, it ties its genes into machine-readable functional units so you're not looking at them in isolation. My opinion, anyway.

A genomic view of the microbiome of coral reef demosponges AbstractSponges underpin the productivity of coral reefs, yet few of their microbial symbionts have been functionally characterised. Here we present an ana

Like here our criteria was that >70% of genes in a an enriched pathway had to be present, including the "key" genes required for proper functioning. One enriched gene says little, but we try to make it as unlikely that we're focusing on statistical accidents.

academic.oup.com/ismej/articl...

Yeah, EnrichM just had something like a Bonferoni correction applied to the p-values. And like any gene/based analysis like that, I’d only consider it valid if, say, >50% of genes in a module or pathway were enriched. We don’t come up with big stories based on a single gene.

In EnrichM, you could supply two sets of MAGs, it would annotate with KEGG/PFAM/CAZY, and give outputs showing which KOs and KEGG modules, PFAMS, and CAZY groups were statistically enriched in one set over another, giving a p-value for each KO/PFAM/Module and fold enrichment between the two groups.

Curious, if someone wanted to annotate a large set of MAGs with KEGG, PFAM, CAZY, and then do some statistical comparison, how would you do that these days?

Back in the day I loved EnrichM, but the backend databases for KEGG, etc, are woefully out of date now.

To fill in a bit, the reason I think The Expanse is the GOAT is because it’s got an epic story arc, super interesting world building, but it’s not dense, has great character development. I get exhausted by sci-fi that’s trying to look smart with the sci but don’t have characters I truly care about.

Bringing the uncultivated microbial majority of freshwater ecosystems into culture - Nature Communications A large fraction of aquatic bacteria remains uncultured. Here, the authors cultivated 627 strains of abundant freshwater bacteria from 14 European lakes, thus generating a collection that includes many previously uncultured, oligotrophic bacteria that may serve as model organisms.

⚠️New paper ➡️ isolating >600 freshwater microbes using dilution-to-extinction and low-nutrient, lake-mimicking media.

🤯Expands cultured diversity to ~70 % of detected genera and enables physiological studies of streamlined oligotrophs.

🔗 www.nature.com/articles/s41...
#protistsonsky

Team, I need a new book/series to read. I’m partial to Sci-Fi or fantasy, points for anything that “changed your life” or got you super reved.

Books/authors I love:
The Expanse series (personal GOAT)
The Bobiverse
Anything Kurt Vonnegut or Neil Gaiman

Already simul-reading 4 non-fiction.

Giant DNA viruses encode a hallmark translation initiation complex of eukaryotic life In contrast to living organisms, viruses were long thought to lack protein synthesis machinery and instead depend on host factors to translate viral transcripts. Here, we discover that giant DNA virus...

Historically, viruses were thought to primarily use host cell's translational machinery. New work from @harvardcellbio.bsky.social faculty Amy Lee reveals that a giant DNA virus encodes its own IF4F initiation complex, suggesting an unexpected evolutionary innovation. www.biorxiv.org/content/10.1...

Systematic comparative benchmarking of computational methods for the detection of transposable elements in long-read sequencing data www.biorxiv.org/content/10.1... 🧬🖥️🧪

Efficient sequence alignment against millions of prokaryotic genomes with LexicMap - Nature Biotechnology LexicMap uses a fixed set of probes to efficiently query gene sequences for fast and low-memory alignment.

Interested in aligning your long sequences or small genomes against huge reference databases containing millions of prokaryotic genomes ? A new tool has been released that can do this efficiently - LexicMap - www.nature.com/articles/s41...

They’re coming y’all…the megaplasmids are coming

(I say this in a slightly frustrated manner as someone who has been thoroughly unsuccessful in getting funding to study megaplasmids for the better part of 15 years)