And it didn't stop when the balance reached 0 (already $-300)! 🧪🧬🖥️✨

Hi everyone, heads up that openAI API monthly limits appear to not work! We got an email that our jobs passed the limit and they're still going (already $900 with a $400 limit). gpt-5_1 also costs more than we expected - need more real-time math!
🧪🧬🖥️✨

Baby-to-baby strain transmission shapes the developing gut microbiome - Nature A metagenomic survey of babies attending the first year of nursery detected extensive baby-to-baby microbial strain transmission, pointing to social interactions in infancy as crucial drivers of infan...

Babies sharing microbes is a leading factor in gut development! A study across infants, educators, and families finds extensive transmission between babies during the first nursery year - early social interactions greatly shape their microbiome!
www.nature.com/articles/s41...
🧪🦠 #microbiomesky

It is with deep sadness that we share the passing of Professor Yehuda Cohen, a founding figure of the International Society for Microbial Ecology and one of the architects of modern microbial ecology.
isme-microbes.org/in-memoriam-...

🤣exactly

The Isomorphic Labs Drug Design Engine unlocks a new frontier beyond AlphaFold - Isomorphic Labs Today, we are excited to share an update on our progress towards a new frontier of drug design. We have unlocked a new paradigm of predictive accuracy in understanding our biomolecular world, allowing...

AlphaFold shifts to Closed Science? AF4, from DeepMind drug spin-off, outperforms AF3 by 2.3x and Boltz-2 by 20x in antibody-antigen predictions. Strictly proprietary + technical report lacks details for outside scientists to replicate the results
🧪🧬🖥️🦠✨
www.isomorphiclabs.com/articles/the...

Bluesky Map Interactive map of 3.4 million Bluesky users, visualised by their follower pattern.

I made a map of 3.4 million Bluesky users - see if you can find yourself!

bluesky-map.theo.io

I've seen some similar projects, but IMO this seems to better capture some of the fine-grained detail

Advancing regulatory variant effect prediction with AlphaGenome - Nature AlphaGenome, a deep learning model that inputs 1-Mb DNA sequence to predict functional genomic tracks at single-base resolution across diverse modalities, outperforms existing models in variant effect...

AlphaGenome is out! Input 1 Mb DNA -> predict gene expression, transcription initiation, chromatin accessibility, histone modifications, transcription factor binding, chromatin contact maps, ..., up to single-base-pair resolution! Trained on human and mouse data 🧪🧬🖥️🦠✨
www.nature.com/articles/s41...

Fecal microbiota transplantation plus immunotherapy in non-small cell lung cancer and melanoma: the phase 2 FMT-LUMINate trial - Nature Medicine In a phase 2 trial evaluating healthy donor fecal microbial transplantation plus either anti-PD-1 in patients with non-small cell lung cancer or anti-PD-1 and anti-CTLA-4 in patients with melanoma, en...

FMT before immune checkpoint inhibitors improves response in NSCLC and melanoma. E.citroniae, E.lavalensis, C.innocuum persist in non-responders: suppress ICI efficacy? FMT works by depleting harmful taxa. Next step: targeted MO depletion?
#microsky #microbiomesky 🧪🦠
www.nature.com/articles/s41...

Exploring functional insights into the human gut microbiome via the structural proteome Liu et al. develop a structure database of human gut microbial proteins and a structure-aware AI model for remote homolog detection. They demonstrate the power of structure-guided approach in discover...

A structural database of ~2.7M predicted human gut microbial proteins + an alignment-free method by structure-encoded protein language models for fast and sensitive detection of remote homologs - really cool!
#microsky 🧪🧬🖥️🦠✨
www.cell.com/cell-host-mi...

A related study from Nandita’s group on gene-specific sweeps - a great idea to compare non-synonymous linkage disequilibrium (LD) vs synonymous LD! Gut microbiome show pervasive selective sweeps in carbohydrate metabolism - diet adaptation! #microsky #microbiomesky 🧪🦠
www.nature.com/articles/s41...

In remembrance of Peer Bork  | EMBL EMBL and its community are deeply saddened by the death of Peer Bork, the organisation’s Interim Director General.

very sad news. Peer Bork was one of the leaders of our field, a wonderful scientist, and he's much too young to be gone. www.embl.org/news/embl-an...

Peer Bork on top of Zugspitze, ready to race down

Peer Bork, Jan Korbel, others at Sonnenkar ski lift at Zugspitze

Après-Ski scientific sessions from 16:00 till very late.

For many years, Peer Bork organised Bio-IT retreats to bring together the bioinformaticians across EMBL, in Garmisch-Partenkirchen and later Feldberg. Days out on the mountain followed by forward-looking and meta-science oriented sessions in the hut that fueled many projects and careers.

So happy to host Dr. Xiaoqian (Annie) Yu - soon starting her lab at University of Montreal! She used population genomics and large data analysis to group gut microbiome into ecological units based on sweep events (major revision at Nature)!
www.biorxiv.org/content/10.1... #microsky #microbiomesky 🧪🦠

The specificity and structure of DNA cross-linking by the gut bacterial genotoxin colibactin Accumulating evidence has connected the chemically unstable, DNA-damaging gut bacterial natural product colibactin to colorectal cancer, including the identification of mutational signatures that are ...

Oops - thanks! Right link here www.science.org/doi/10.1126/... #microsky #microbiomesky 🧪🦠

Structure and specificity of the colibactin-DNA interstrand cross-link. Characterization of DNA damage by the human gut bacterial genotoxin colibactin reveals a preference for alkylation at AT-rich DNA sequences and the importance of an unstable central α-ketoimine in mediating this specificity. These results help to explain the locations of cancer-linked mutations derived from colibactin exposure.

Happy New Year, everyone! First paper to share in 2026: (finally) a mechanistic study of cancer microbiome! pks+ E. coli's colibactin binds AT-rich DNA in minor groove and induces cross-links in colorectal cancer cells #microsky #microbiomesky 🧪🦠
www-science-org.remotexs.ntu.edu.sg/doi/10.1126/...

Bile acids target at STING dimer-dimer interface. (a) HEK293T transfected with pMSCV-hygro-STING were treated with DCA-4 (10 µM) in presence of different concentrations of NVS-STG2 for 30 mins at 37 °C, followed by 365-nm light irradiation. The labeled proteins in cell lysis were reacted with az-biotin and enriched with streptavidin beads (‘pulldown’). Equal loading was validated by immunoblotting of input samples before enrichment (‘input’). (b) Quantification of binding efficiency in A by grayscale analysis (pull down/input) (n = 5 - 6 biologically independent samples.). Data were analyzed by unpaired two-sided t-test. (c) High-resolution cryo-EM map of the two STING dimer with highlight of dimer-dimer interfac molecular density. (d-f) HEK293T transfected with pMSCV-hygro-STING (WT and mutations), IFN-Beta_pGL3 and pRL-TK control plasmid were treated with diABZI (20 nM) in presence or absence of DCA (n = 5 biologically independent samples). A two-way ANOVA followed by Sidak’s multiple comparisons test was used to calculate adjusted P values. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Prof. Howard Hang (Scripps) shared a great story on forward and reverse chemical microbiology using model organisms during his visit! A sneak peek: microbiota-derived secondary bile acids promote STING activation and antitumor activity #microsky #microbiomesky 🧪🦠
www.biorxiv.org/content/10.1...

Our first year PhD student, Qiwen, presenting our work on biology-guided AI models to predict bacteria protein DNA binding, on BUG seminars! Great job! Thanks, @cwhitch.bsky.social for organizing this! 🧪🧬🖥️🦠✨

Thanks!

Also, our seq alignment tool Mapper can infer ancestor references and reduce the penalty for a basepair mismatching a ref but matching its ancestor.

The main idea is about using phylogeny to give feedback to users about which parts of the reference tree lack important genomes.

A secondary idea is that our seq alignment tool Mapper can infer ancestor references and reduce the penalty for a basepair mismatching a ref but matching its ancestor.

YAY this tool might be able to use that information to give feedback about which part of the reference tree lack important genomes!

YAY thanks! maybe we can even choose several least bad references and/or a pangenome.

Reference genome choice compromises population genetic analyses Using a species-specific reference genome significantly improves the accuracy of population genetic analyses, as exemplified by mapping gray fox sequence data to genomes of several canid species. Mapp...

Reference genome choice compromises population genetic analyses - a problem for microbes too! Our lab is designing a tool to automatically choose references for input samples using features like phylogeny of reads to existing references. Any thoughts🤩? 🧪🦠🧬🖥️ #microsky
www.cell.com/cell/abstrac...

That's awesome - will share with my people :) Thank you, Sean!

Using leave-one-out cross-validation to calculate metrics such as AUC and R^2 creates bias! This can be fixed by removing one of each class in the meanwhile to maintain the training data distribution - great work by Tal and his team 🧪🧬🖥️

My pleasure to host Dr. Klas Udekwu's seminar on *Dissecting community alteration: a control theory–based approach to microbial community modulation* with @briefutopia.bsky.social Cool math modeling to guide sequential antibiotic or nutrient treatment to manipulate microbiomes! 🧪 🧬 🖥️ #microsky

We're at 735 registrants and counting! 🎉🦠

If you haven't registered yet, don't miss out on this exciting (free) symposium on gut microbial metabolites and their impacts on host health.

Please share with your networks. 🔗👇

isbscience.org/events/2025-...

Boltz-2: an AI model approaching free-energy perturbation accuracy for small-molecule–protein binding affinity!
It builds on Boltz-1, an open-source model from MIT that predicts the 3D structures of biomolecular complexes with AlphaFold3-level accuracy 🧪✨ 🧬 & 🖥️
www.biorxiv.org/content/10.1...

A great pleasure to speak at the SCELSE retreat today! Sharing how to uncover diverse microbial mechanisms via AI-powered computation: regulatory network evolution, environmental sensing, phage defence, and host–microbiome interactions 🧪 🧬 🖥️ ✨ @scelse.bsky.social #microsky #microbiomesky