Thanks for sharing! One caveat I noticed is that all results are based on the Zeevi 2015 dataset with ~2.5 Gb from HiSeq2500 (2x100bp) per sample on average. "Modern" datasets are more in the range of 10 Gb with longer reads - this will have huge impact on MAGs, I suspect!
The image shows a woman smiling and wearing a black jacket against a purple background. Text: #WomenInScience. Prof. Dr. Mathilde Poyet. Professor for Intestinal Microbiology
Quote Mathilde Poyet: My research focuses on developing advanced methodologies to deeply investigate undercharacterized human-associated bacterial species and expanding our understanding of the human gut microbiome’s diversity, evolution, and impact on health across globally diverse populations.
Quote Mathilde Poyet: I enjoy working in science because it combines curiosity, teamwork, and openness, and gives me the opportunity to build inclusive collaborations across cultures and disciplines, while discovering something new every day.
To mark the International Day of Women and Girls in Science #11February, we'll be introducing several female scientists from Kiel University over the next few days.
#WomenInScience #ConnectingHorizons
Today, among others: Prof. Dr. Mathilde Poyet
@mmmicrobiomelab.bsky.social
Amazing. Bypass model safeguards and get helpful responses about nukes, bioweapons, cyberattacks, etc., simply by putting the adversarial prompt in a poem.
Adversarial Poetry as a Universal Single-Turn Jailbreak Mechanism in Large Language Models arxiv.org/abs/2511.153...
I was fortunate to be one of the scientists to analyse this fantastic resource generated by the GMbC @globalmicrobiome.bsky.social and I am extremely thrilled to see these articles out on @biorxivpreprint.bsky.social
#microsky 🦠
💡 Takeaway: In children, enterotypes aren’t innate—they’re ecological outcomes of postnatal microbial colonization and diet. Prevotella-rich states are possible, but take longer to emerge. 🧬🌱 7/n
📄 Link to paper: doi.org/10.1038/s415...
🚸 Weaning as a switch: Mutation rates in gut microbes spiked after weaning, especially in genes for carb metabolism—a likely response to dietary transition. This may help explain enterotype restructuring. 🍞🧪 6/n
🧬 Strain-level perspective: The team reconstructed nearly 4,000 MAGs, tracking which strains persisted or were replaced. Bacteroides strains were more stable, while Prevotella appeared later and were less persistent. ⏳🧫 5/n
🧠 Ecological insight: The Bacteroides–Prevotella divide mirrors adult enterotype structure. But in kids, these configurations are late-emerging, dynamic, and not fixed, challenging static interpretations of enterotypes. ⚖️📉 4/n
🔄 Prevotella dynamics: While less common, Prevotella-dominated states emerged in some children only after age 4. These were mutually exclusive with Bacteroides and often linked to specific dietary contexts. 🥕🦠 3/n
📊 Enterotype emergence: Most infants begin with a low-Bacteroides profile. After weaning, many transition into adult-like enterotypes—primarily Bacteroides-dominated, but also some Prevotella-rich configurations. 🔁🥄 2/n
Sawhney et al. followed 52 infants from birth to age 8, uncovering how gut microbiomes shift, stabilize, and evolve. A key finding: the emergence and timing of Bacteroides- and Prevotella-dominated enterotypes. 👶🧬 1/n
Yes, same! Great piece with some very quotable lines: "Science requires us to be open: we constantly need to follow the data, to branch out in new directions. […] Openness allows creative endeavors to evolve."
🔥New Night Science paper!!
Discovery happens when your initial plans fall apart but it requires you to have a particular mindset: it's not extraverted, orderly, neurotic or agreeable that's the most important – discovery requires an OPENNESS to new ideas and unexpected insights.
🌍 Implications: This model shifts how we view human origins—from panmictic simplicity to structured complexity with implications for selection, ancestry, and interaction with archaic hominins. 🧬📜 6/n
📄 Link to paper: www.nature.com/articles/s41...
🧠 Functional Impact: Regions enriched for minor (B) ancestry include genes tied to neuronal functions, while regions depleted in B ancestry involve immune responses, hinting at selective pressures post-admixture. 🧠🔬 5/n
💡 Link to Archaic Humans: The majority (A) lineage shows closer genetic ties to Neanderthals and Denisovans, suggesting A was ancestral to archaic humans, while B’s contribution was more distant and selected against. 🧬🦴 4/n
🦠 Ancestral Structure: After splitting, population A experienced a bottleneck, while B remained larger. Selection appears to have acted against B’s genetic contribution in humans today, especially near coding regions. ⚔️🧬 3/n
🔑 Key Idea: Rather than a single ancestral population, modern humans trace their lineage to deeply structured populations. Cobraa distinguishes these using coalescent patterns in genome sequences. 🧬📊 2/n
Not microbiome, but not less interesting: Cousins et al. introduce cobraa, a model revealing that all modern humans descend from two ancestral populations that split ~1.5 million years ago and later admixed ~300,000 years ago. This 80:20% ancestry mix challenges the idea of a panmictic origin 🌍🧬 1/n
Big news: we are setting up a new non-profit organization to run bioRxiv and medRxiv. It's called openRxiv [no it's not a new preprint server; it's dedicated organization to oversee the servers] openrxiv.org 1/n
👇Four days left to apply! (Friday 14th of March)👇
Good point. They mention Dorea longicatena as another potential candidate to reduce C.diff infection due to its similar capacity for proline fermentation / Stickland fermentation.
Excited to share our latest work on the gut microbiome in inflammatory bowel disease patients. This is the result of hard work across generations of trainees in the Donia Lab, in collaboration with amazing scientists (Nobuhiko Kamada and Lea Ann Chen). www.cell.com/cell-host-mi...
💡 Implications: This study moves beyond FMTs, showing how a rationally designed microbial therapy can replace donor stool with precisely defined consortia, improving safety and reproducibility. 🎯💊 6/n
📄 Link to paper: www.sciencedirect.com/science/arti...
🐭 P. anaerobius Alone Protects: Mono-colonization with P. anaerobius provided protection equivalent to human FMT in a gnotobiotic mouse model, suggesting potential for a single-strain therapy. 🏥🦠 5/n
🦠 Proline Fermentation is Key: Rather than bile acid metabolism, proline-fermenting bacteria, especially Peptostreptococcus anaerobius, were necessary and sufficient for suppressing C. difficile via nutrient competition. 🔄🥩 4/n
🤖 Machine Learning Design: The team analyzed 12 human microbiome studies, identifying microbes negatively associated with C. difficile. These predictive signatures guided the design of sFMT1. 📊🛠️ 3/n
🔑 Key Idea: Rather than relying on human donor FMTs, this study shows that a designed microbiome can provide targeted protection, identifying specific bacterial functions responsible for pathogen suppression. 🧬🦠 2/n
Tian et al. designed a synthetic microbiota that suppresses C. difficile infection. Using machine learning, they built a 37-strain synthetic fecal transplant (sFMT1) that successfully inhibited C. difficile in vitro and in animal models. 🦠🛠️ 1/n
Lab mice (domesticated for >100y) have kept gut bacterial strains that codiversified w/ rodents for >25My. Here, authors show genetic drift occured in the lab, leading to loss of microb diversity+more deleterious mutations, impacting microbial fitness in lab vs wt mice
www.nature.com/articles/s41...