Профиль

@jgi.doe.gov's Genome Insider podcast has posted an episode about our paper now too, featuring @simrouxvirus.bsky.social and me: jgi.doe.gov/user-science...

The whole issue of Nature Micro is up now, and look at that gorgeous cover! @emerge-bii.bsky.social co-director Ruth Varner took the picture. www.nature.com/nmicrobiol/v...

Don't let our paper stop you! We are really hoping that the analytical framework we've laid out helps people dive deeper into mobilome analysis in other systems too.

We also need funding for interdiscliplinary teams that can combine deep knowledge of site history, hydrology, biogeochemistry, informatics, modeling and more to not only do the work but also understand it in context and work out how to scale up to the wider world. 23/23

If we want to keep pushing on these questions, we need sustained investment in long-term observation of model ecosystems. These analyses are data- and timepoint-hungry. When you're patching together stopgap funding to keep your team employed, you can't dream up a sampling plan like this. 22/23

We have no idea where biological reality sits between these extremes. Seems like the kind of thing we ought to figure out, right? Because microbial adaptibility affects everything in a changing world: soil health and crop yields, climate, human health, you name it. 21/23

(2) Activity is stochastic, say Poisson-distributed. In a population with 38 MGEs/cell, 1% activity means almost a third of cells have at least one MGE firing. At 66 MGEs/cell, 1% activity means about half of cells have at least one MGE firing, and >14% have 2 or more. 20/23

Extreme scenarios: (1) All that activity is concentrated in a minority of cells that are freaking out and have every MGE active. Huge combinatorial potential in those cells; elsewhere, pretty chill. 19/23

We looked at activity in 3 different ways, capturing 3 different timescales. The most straightforward is expression of MGE recombinase genes. By that measure, ~1% were active at any given time. Seems like very few, until you remember: 38-66 MGEs per genome. 18/23

That is, these wildly abundant elements are poised to affect the adaptibility and activity of pathways that feed into global biogeochemical cycles. How often do they actually move around? 17/23

We could make confident functional predictions for a subset of those, and the answer is: MGEs affect genes doing all the things. Antibiotic resistance, stress response, sure, fine, yes. But >50% of genes annotated affect "domestic" processes involved in being a cell. Notably, nutrient cycling. 16/23

But for a tiny subset of our MGEs, we could identify genes we were highly confident were MGE-affected, either because they were being moved or because they were host genes being interrupted by transposons. 15/23

With our 2.1M hits plus a new bound on how much we're missing, we estimate that microbial genomes in the wild have an average of 38-66 MGEs each. How do they affect their hosts? Most of the time, we can't say: defining MGE boundaries is really hard, and it's even harder in metagenomes. 14/23

Just as you'd expect, short-read-only MAGs recover just 35-60% of the MGE recombinases detectable in hybrid assemblies. Just as you'd expect, it's worst for recombinases that move more and have higher nucleotide diversity. Coverage makes a real difference. 13/23

So @aroneys.bsky.social and @benjwoodcroft.bsky.social benchmarked with 6 samples sequenced with both short and long reads. Among the very high quality MAGs (>95% complete, <5% contam) recovered, they found 17 species-level clusters with members from both hybrid and short-only assemblies. 12/23

Unsurprisingly, most (>5 out of 6) of our hits were in contigs that failed to make it into MAGs. (Again: this was all a crazy idea; MGEs live to thwart us.) We figured we were probably missing even more that never made it into contigs. 11/23

And did we ever find them! 2.1M MGE recombinases, out of 131.5M predicted genes altogether. Which is to say: about 1 in 63 genes in these soil metagenomes was an MGE recombinase. That's a lot! More than 7X as common as the most common family of actual host genes detected (rpoE). 10/23

First question: how many MGEs in our metagenomes? With Matt's postdoc Jiarong Guo taking point, we used an excellent set of HMMs developed with high-quality genomic sequence data in the lab of the late great Peer Bork (doi.org/10.1093/nar/...) to go looking for MGE recombinases in our data. 9/23

Since 2010, the IsoGenie Consortium and its successor EMERGE have been sampling soils from Stordalen Mire, a thawing permafrost peatland in northern Sweden. The complete dataset is pretty mind-blowing. emerge-db.asc.ohio-state.edu 8/23

In fact there just aren't that many datasets out there where this plan would have been merely crazy rather than obviously doomed. But we had one, from a fascinating site, and between @emerge-bii.bsky.social and the VirSoil Consortium we had the team to analyze it. (Photo: Scott Saleska) 7/23