Alex Crits-Christoph

Adapted from Axelsson et al. 2013 Fig 2c: Histogram showing the distribution of diploid amylase copy number in wolf (n=35, blue) and dog (n=136, red). Dogs carry more copies of the starch-digesting gene AMY2B than wolves. Additional copies make dogs better than wolves at digesting starchy foods like grains & vegetables.

Dogs evolved to eat your leftovers! Comparing dog & wolf genomes revealed dogs have up to 30 EXTRA copies of the amylase gene (AMY2B) that helps digest starch. This is a key genomic signature of living alongside humans & table scraps for thousands of years 🐕 www.nature.com/articles/nat... #2026MMM

When you're actually right about an association being causal, but you're way off mark about the mechanism.

It is pretty crazy that claude can often one-shot the prompt: "now do it in C"

How can we tap into the enormous functional diversity of microbes on this planet?

Join us at EMBL Heidelberg (8–10 Dec 2026) to learn about ongoing efforts to decode gene function and organisation in human gut microbes, and present your science.

your response is excellent

Image from their link showing evidence of fabricated references

Holy smoke. What ultimately happened???

"predicts variants it has seen a lot" is not necessarily bad though, right?

"predicting variants seen a lot in similar contexts, relying also on contexts that are less similar but still share some features" is the whole ball-game, right? But TBF some think there's magic beyond that, which is silly

still confused, but I should just read more closely and read the code!

So do you think pLMs still at least have a good understanding of which sites are important? if I change each site to a pro, will the logits for that be lower in helices or near active sites? And worse than ala?

"a naive model that simply predicts the mean (known) fitness value at each site"

Dumb question - so isn't this actually a pretty tough comparison? For most proteins of course, we don't have any fitness data. If pLMs can do as well as mean fitness per site, is that not pretty useful?

New paper showing that much of the apparent success of protein language models in predicting mutational effects is a mirage: These models mostly memorize sites. 1/
www.biorxiv.org/content/10.6...

AF2BIND: predicting small-molecule binding sites using the pair representation of AlphaFold2 Nature Methods - AF2BIND is a logistic regression model trained on AlphaFold2 pair features to predict small-molecule binding-site residues in proteins, without multiple sequence alignments,...

Our paper with @sokrypton.org using AlphaFold2 to predict small-molecule binding sites in proteins is now out in Nature Methods! 🧵
rdcu.be/e7SnX
www.nature.com/articles/s41...

this is awesome work!

Automated genome mining predicts structural diversity and taxonomic distribution of peptide metallophores across bacteria Automated detection of metallophore biosynthesis reveals that metal-chelating non-ribosomal peptides are widespread, chemically diverse, and deeply rooted in bacterial evolution.

Are you interested in how to predict functions of natural product biosynthetic gene clusters (BGCs) and their products? And/or do you love metallophores and would love to identify their producers in microbiomes? Check out @zachreitz.bsky.social 's new paper!
elifesciences.org/articles/109... 1/n

How Congress can restore the independence of US science Members must go beyond reinstating US government research spending and re-establish decentralized governance at the National Institutes of Health and other agencies.

The most impt change at #NIH and to US science this year is bigger than grant cancellations— it’s how the agency is governed.

For 75 years NIH has been largely independent of presidential control. That’s changed this year. New piece from me and @nataliebaviles.bsky.social in @nature.com
🧪

Weird day for microbiology

Live Updates: Strikes Launched Across Middle East, With No Compromise in Sight

President Trump on Saturday blamed Iran for an airstrike that hit an Iranian elementary school on February 28, killings scores of children. An analysis by The New York Times indicates that the school was most likely hit by an American airstrike. Follow live updates.

Massachusetts Runs on Research Funding: Support the DRIVE Act The DRIVE Act, part of an initiative proposed by Governor Maura Healey, fights back against the federal government's attacks on science by ensuring continued investment in critical life-saving researc...

and the links:
region9a.uaw.org/news/driveact
www.mafamiliesforvaccines.org/actioncenter
www.aaup.org/event/nation...

Stand up For Science rally in Boston, March 7 2026

Great rally in Boston today! Here are some actions anyone can take:

-- support the DRIVE Initiative, which would invest $400M in MA research+higher education

-- support H.2554, which will close irresponsible loopholes in vaccine requirements for children in MA

-- Protest ICE and war on March 28

A phylogenetic tree of insects is shown annotating the presence or absence of a an antimicrobial peptide gene across winged insects

Various phylogenetic secondary loss events are mapped to a tree of insects to explain the parsimony calculations necessary to explain the diversity of insect Drosomycin antimicrobial peptide genes

Antimicrobial peptides (AMPs) are key defence molecules of the innate immune system of plants and animals. Understanding the evolutionary origins of AMPs can help to explain how immune systems acquire novelty and vary in their defensive capabilities. However, AMPs evolve rapidly, and so the origins of similar AMPs across organisms is often unclear. Furthermore, false negatives due to low search sensitivity are common and can hinder confident annotations about true absences. Due to these difficulties, understanding whether similar AMP genes found in diverse organisms represent ancestral molecules or evolutionary novelties has been challenging. In this report, we present evidence of horizontal gene transfer (HGT) of the antifungal peptide gene Drosomycin across insects. We show that in Diptera, the presence of Drosomycin is restricted to the Melanogaster group and additionally the distant relative Drosophila busckii. We go on to recover Drosomycin genes in cockroaches (Blattodea), mantises (Mantodea), one katydid (Orthoptera), various beetles (Coleoptera), and a recently acquired pseudogenized Drosomycin locus in Liposcelis booklice (Psocodea), but no other insects. Explaining this diversity through shared ancestry requires at least 50 independent loss events, or just seven HGT events. Previous studies have suggested that similar AMPs found across divergent species reflect conservation from a common ancestor, or due to their small size, that they arose via convergent evolution resulting from pathogen-imposed selection. Our findings suggest horizontal gene transfer can be responsible for the presence of some AMP genes found scattered across the tree of life. By presenting a mechanism through which immune systems can acquire novelty, our study also suggests a possible explanation for certain lineage-specific competencies for defence against infectious disease. While loss of AMP genes is common in certain lineages, here we suggest gain of AMPs can occur just as suddenly.

Pleased to finally share this fun collab that began at #Ento23

@cedricaumont.bsky.social presented & I had seen NCBI annotated some cockroach genomes as "contaminated." Turns out NCBI & I were wrong (much more fun).

Horizontal transfer of an #AntimicrobialPeptide across insects
bit.ly/DrsHGT

1/🧵

The data also reinforce a message that should be clear: American leadership in science and medical research requires far more than getting across the line on top-line budget negotiations and will require continued vigilance to track how HHS treats applications and existing awards.

Deep untargeted wastewater metagenomic sequencing from sewersheds across the United States Wastewater monitoring enables non-invasive, population-scale tracking of community infections independent of healthcare-seeking behavior and clinical diagnosis. Metagenomic sequencing extends this cap...

Happy to report that we submitted a new manuscript this week.

The manuscript is abour our work untargeting wastewater sequencing as a technique for monitoring viral pathogens from wastewater. @lennijusten.bsky.social
1/
medrxiv.org/cgi/content/...

Don't panic, Organize! It was great to be able to represent @nihfellowsunited.bsky.social today at the Take Back Our Science Rally in DC to speak out against the attacks on our union and for scientist workers rights!

The problem with this poll is they polled the general public. But the general public is a well known echo chamber!

Should have polled X, which is well known for not being an echo chamber.

https://www.annenbergpublicpolicycenter.org/stark-divide-americans-more-confident-in-career-scientists-at-u-s-health-agencies-than-leaders/

A quick email before I head of to get ready for the @standupforscience.bsky.social rally.

1/3

Great news for the toxic PI that wants RNA seq data analysed in under a minute.

SUFS rallies this weekend will be so lit that Vinay has preemptively surrendered

Excited to share great work from @jingqian.bsky.social , now a second year graduate student in my lab, from research she conducted at her previous position at Shanghai Jiao Tong University!

Would have been cool to see comparisons to the SARS2-related spillovers in pangolins, and to the SARS-CoV-2 spillovers to white tailed deer!

In each case we know that there were spillovers, and testing each would show how sensitive these methods are to known spillovers of different sizes

MKado: a toolkit for McDonald-Kreitman tests of natural selection https://www.biorxiv.org/content/10.64898/2026.03.02.709122v1