Steven Robbins

The Microflora Danica atlas of Danish environmental microbiomes - Nature Microflora Danica—an atlas of Danish environmental microbiomes—reveals that although human-disturbed habitats have high alpha diversity, species reoccur, revealing hidden homogeneity.

Microflora Danica: What can you learn from collecting and sequencing 10,000+ samples from a single country? Check out our new paper in @nature.com to find out. Incredible work led by Caitlin Singleton, Thomas B. N. Jensen, and Mads Albertsen from @aau.dk. 🦠🧫🧬
www.nature.com/articles/s41...

Meta-virus resource (MetaVR): expanding the frontiers of viral diversity with 24 million uncultivated virus genomes Abstract. Viruses are ubiquitous in all environments and impact host metabolism, evolution, and ecology, although our knowledge of their biodiversity is st

🦠🧪🧬🚨 New paper and database alert: the new IMG/VR release is now MetaVR ! We have a new website - meta-virome.org - with quick search capabilities for the >24M viruses, >12M vOTUs, and >42M protein clusters (including >790k with predicted structures !). academic.oup.com/nar/advance-...

Hologenomic insights into the molecular adaptation of deep-sea coral Bathypathes pseudoalternata Wei et al. reveal how deep-sea corals partner with symbionts in the extreme environment using the hologenome of Bathypathes pseudoalternata. A division of labor shows that the host provides a protecte...

Symbiont-coral partnerships

Hologenome of B. pseudoalternata reveals a division of labor between deep-sea coral & symbionts. Host provides habitat & manages immunity, while symbionts recycle waste, supply nutrients, resist viruses & produce antioxidants
www.cell.com/cell-host-mi...

Enormous milestone. All thanks to vaccinations!!!

Long-read metagenomics reveals phage dynamics in the human gut microbiome - Nature Complex prophage integration dynamics, including low-level induction, cross-family host range and transposase-mediated mobilization, challenge existing paradigms and deepen our understanding of phage–...

Long read Metagenomics, #phage and #prophage in the gut by Ami Bhatt's group. Beautiful data showing changes in phages over two years

#phagesky

www.nature.com/articles/s41...

Thanks Greg. Quarto is cool, but I don't really need to render the code or make it overly fancy. I really just want a notebook to markdown-style application to document what i'm doing, with the code that i'm using to execute things on the server.

Hey Team! What do you think are the best free digital notebooks these days.

I've used:
Evernote--great, but now it's paid

TiddlyWiki wasn't bad, but got rid of scrollable code blocks

Obsidian--basic, and I don't like not being able to scroll my code blocks like OG Tiddlywiki.

Suggestions?

📌

Thank you! Can I ask, do you find it better than ChatGPT5 at the moment?

Super helpful. Thanks Sebastian!

Thanks man! Granted, what I saw with my eyes was a much fainter version of these, and these were points of the night that were strongest. Most of the night you couldn't see much. So these are definitely the enhanced highlight reel.

proGenomes4: providing 2 million accurately and consistently annotated high-quality prokaryotic genomes Abstract. The pervasive availability of publicly available microbial genomes has opened many new avenues for microbiology research, yet it also demands rob

proGenomes4: providing 2 million accurately and consistently annotated high-quality prokaryotic genomes academic.oup.com/nar/advance-... #jcampubs

A Grammar of Graphics for Comparative Genomics An extension of ggplot2 for creating complex genomic maps. It builds on the power of ggplot2 and tidyverse adding new ggplot2-style geoms & positions and dplyr-style verbs to manipulate the…

gggenomes: A Grammar of Graphics for Comparative Genomics thackl.github.io/gggenomes/ #Rstats

Them too, for sure!

For mmseqs, I would think assembled contigs make more sense, but I don't know who i'd trust more than Titus to know where and when to trust an assembly.

That's the nice thing about long reads. The reads themselves are the primary data, and long enough to be their own evidence. It's not underrated. The issue now is it's kind of easy at the moment to find skeletons hiding in the assembly/binning closets, we just couldn't see them before.

Just from experience, it's the way i'd go with, say, long read data or contigs with at least a few genes on them. If there's half a gene on each "read," LCA classification won't go that far...maybe at least to domain, though, which is the goal here. Curious to see if it works.

If they're like 150bp reads, you're not going to get good classification though. It'll get better the more genes are on those contigs. There's really just no good way to classify short reads that's reliable, as far as i'm aware.

Ah ok, so by this I think you mean you have a subset of reads in a MAG that you want to classify as euk or prok. If you have something like a full gene or multiple genes, LCA might be what you want. Works better on assembled contigs since it's looking at what multiple genes on the sequence blast to.

Yup...I would just say there's no good answer to this. Best I can think of is LCA-assignment in mmseqs. At least you might get a "most" likely it's in this domain of life. I tend to stay away from pure classifiers for things like this except in special cases.

Those suggesting below to look for ribosomal genes, etc, work for the first case, not for the second. If they're random contigs you want to classify, the only good way to really do it is something like Lowest common ancestor-assignment as implemented by mmseqs.

Need a bit more info on what you mean. By "binned into a bacterial genome," do you mean that you want to make sure the MAG is prokaryote and not euk, or a subset of reads that got binned into a MAG are not euk. The first is easier, the second probably doesn't have a good answer.

I think he’s saying he’s got some random genes in a bin, right? For graftM or singlem to work they’d have to be a core marker gene or have a specifically curated graftM package to classify with. Neither seem likely?

ECR connection: Meet Meren when you need to A means for ECRs to get advice from a senior scientist outside of their support network

Starting this week, I set aside one hour each week to meet ECRs outside my group who want to discuss career development, mentorship, or any non-technical professional questions.

Here is a blog that explains my motivation for this and how to schedule a meeting:

merenlab.org/2025/11/16/E...

“My magic beans can work miracles, they’re different from all of the fake magic beans out there” has been the sales pitch of charlatans since time immemorial.

My colleague saw no reads mapping to some regions and other reads all truncating all at the same bp position like their ends didn't map anywhere.

But really, first i'd just see if your 5S is really there by blasting some closely related species' 5S to it. Then go down a rabbit hole if it's there.

So, my (probs incomplete) understanding of minimap is it's producing HSPs, best alignments of parts of each read, not trying to align the whole read. What we wanted to see with our Hifiasm misassemblies was whether reads spanned a whole region, and they didn't--i.e. they didn't support the assembly.

@jung-gt.bsky.social I know it’s academic, but it doesn’t *have* to be made up. These companies make crazy profits margins. United healthcare has a 22-26% profit margin. It’s not like they’d go under without the subsidies. They’re choosing to extort their customers because healthcare is necessary.

Hence the mauve alignment to see if things look fishy.

But could also just be barnap not picking it up.

I’ve seen some recent HiFiasm isolate data from a collaborator that produced some confusing misassembly, like duplicated sequences not present in a separate Flye assembly, stretches of sequence with no reads mapping to them.