...and its authored by Sever, et al. - just one R shy of making it a meta^3

(3) most surprisingly❗️there was *not* a correlation between population size effected, and rate of power restoration. This can be seen here where Wellfleet (green), Yarmouth (gold) and Barnstable (red) have highly time-correlated % recovery slopes, despite having wildly different number of households

Some takeaways unique to the historical trend data: (1) most of the repairs happen during the day and early evening (2) power restoration was not always a linear (*er, monotonically increasing) process - especially on the Outer Cape

Cape Cod Power Outage Dashboard showing a “Power Restoration by Town” line chart (y-axis: percent of customers with power, 0–100%; x-axis: time from Feb 25 to Feb 27, 2026). Many towns appear as faint colored step lines, while three selected towns are emphasized with thick lines: green (Dennis) climbs from roughly 25–30% to ~100% by late Feb 26; orange (Brewster) rises more gradually from ~5% to about 90–95% by Feb 27; yellow (Eastham) stays near 0–20% for much of Feb 25–26, then jumps sharply to near 100% late Feb 26. A small inset map of Cape Cod at top right highlights the selected towns, and the dashboard notes it was last updated at 8:24 AM.

Shoutout to the lineworker crews restoring power on #CapeCod after the #blizzard last week - still a few thousand without power but the cleanup and repair is a massive endeavor. I built a dashboard to track the progress (MEMA only shows current conditions)

The cool thing about this transformer model based approach (over diffusion, eg) is that they were able to abstract away the issues diffusion models face with intrinsically disordered regions, engineering IDRs without knowing structure. Very elegant & results-focused solution

🧪🧬🔬💻👾

Accelerating life sciences research OpenAI and Retro Biosciences achieve 50x increase in expressing stem cell reprogramming markers.

Fine-tuning GPT-4o mini on amino acid sequence with biological significance text + structural features mapping enables *function-directed* protein design!

Folks @ Retro achieved 50x improvement on iPSC generation using engineered versions of OSKM over wild-type!

🧪🧬💻👾

openai.com/index/accele...

Big step forward in phenotype-directed drug discovery:

By learning the latent representation of phenotype space, they speed up and directly learn the phenotype transition --> perturbagen mapping.

Super cool!!

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

#AI #drugdiscovery #cancer

A competitive regulatory mechanism of the Chd1 remodeler is integral to distorting nucleosomal DNA - Nature Structural & Molecular Biology Nodelman, Folkwein et al. define a regulatory region in Chd1 containing adjacent inhibitor and activator elements that compete for binding to the remodeler ATPase. The competition between these elemen...

New collaborative paper between JPArmache and Bowman (@bowmanlab-jhu.bsky.social) labs show how the yeast CHD1 chromatin remodeler depends on activator elements to distort nucleosomal DNA. This explains how the NegC inhibitor blocks activity. www.nature.com/articles/s41...

Congrats, JP! It's a really intriguing model for regulation over early ATP-dependent remodeling events! I wonder if there are PTMs that regulate transition or even more distal events like TF binding that could promote it... Very nice story!

Beautiful foggy morning run on Martha’s Vineyard… a little time away from the lab goes a long way to help keep me focused back in the lab!

Nice to see my colleague and friend Sangwoo’s PhD work out and about! Bravo, Dr Park!! 🎉

3b/ This underscores the double-edged sword of #aging intervention: cGAS-activation of senescence helps prevent oncogenic transformation, but too much cellular senescence means runaway inflammation. Maybe we should treat the underlying genomic instability and distal inflammation separately?

🧪

3a/ Chen highlighted the age-related increase in cGAMP and mused about the use of cGAS inhibitors in inflammation associated with #aging. But should we really be thinking about inhibiting the proximal sensor of age-related genomic instability?

🧪

2/ Nuclear cGAS seems to be very strongly associated with nucleosomal DNA - does it have a role on #chromatin or is it just sequestration?

🧪

Kaplan-Meyer curves demonstrating cGas haploinsufficency nearly fully rescues Trex1 complete loss. Source: Gao et al, PNAS 2105 https://doi.org/10.1073/pnas.1516465112

1/ Why does haploinsufficency of cGAS nearly completely rescue Trex1-loss? Chen mentioned that monoalellic loss of STING is also sufficient, so can’t just be cGAMP levels, right?

Source: Gao et al, PNAS 2105 doi.org/10.1073/pnas...

🧪

Zhijian J. Chen presenting an introduction slide of the cGAS-STING pathway in an auditorium at Harvard Medical School

Great round up of cGAS-STING mediated dsDNA sensing with Zhijian Chen - lots to think about!

#HarvardImmunology 🧪

Hahahaha, I was looking for the same!They use AlphaFold for structure prediction of their hit, I think that’s it… such silly clickbait for an actually quote interesting paper that has almost nothing to do with AI

Really nice seminar by Liling Wan here @DFCI! "Chromatin Regulation in Cancer: Molecular Insights and Therapeutic Opportunities"

Many of the DFCI Seminars in Oncology are recorded and open to the public, enjoy!

seminarsinoncology.dana-farber.org/2025.html

🧪

This is a report of correlational findings in an observational study. If you make many measurements (as they did) you will find some correlate. Nothing about this is 'critical'. It might constitute preliminary data in support of doing an actual experiment to validate any possible role, likely none

⚠️ Dangerous article alert ⚠️

This article makes huge, speculative leaps about treatment for a disease based only on correlative findings.

Science communication matters 🧪

NIH funding delivers exponential economic returns — Harvard Gazette Report finds all 50 states reap gains in patient health, job creation, research resources, business development.

Actuarial studies aren’t going to convince anyone who wants to defund the #NIH but in terms of ROI that seems like a pretty solid multiple, not to mention the benefits to life and health! 🧪

St. Laurent Lab - Join We're hiring! Reach out if you know folks looking for postdoc or research associate positions The St. Laurent Lab at the Brigham and Women’s Hospital and Dana Farber Brigham Cancer Center in Boston i...

Interested in joining an exciting new translational gyn/onc research group?! Check out my colleague Jess St. Laurent’s group at MGB and Harvard Medical School!

I cannot recommend Dr St Laurent as a mentor highly enough, she is brilliant, passionate, kind & dedicated! 🧪

www.stlaurentlab.org/join

Genome-coverage single-cell histone modifications for embryo lineage tracing www.nature.com/articles/s4...

In the most recent episode of the Epigenetics Podcast, we talked with Geeta Narlikar from UCSF about her work on chromatin remodeling, Heterochromatin Protein 1, and the molecular mechanisms that influence the genome. #podcast #epigenetics

Listen here: buff.ly/4ibPPUD

Scientists sitting around a table looking at a presentation of B-cell differentiation

It turns out I am a stage of lymphocyte differentiation, apparently a CLP gives rise to a fully differentiated @aido!

File under: Things I learn in lab meetings

Is anyone aware of an exhaustive list of all human histone PTMs identified?

Something like CST’s, except (1) continuously updated and (2) describing the actual modification (e.g. ‘symmetric dimethylation’, instead of ‘methylation’)…

#epigenetics #biochemisty 🧬🧪

Nucleosomal asymmetry shapes histone mark binding and promotes poising at bivalent domains Promoters of developmental genes in embryonic stem cells (ESCs) are marked by histone H3 lysine 4 trimethylation (H3K4me3) and H3K27me3 in an asymmetr…

www.sciencedirect.com/science/arti... and its open! Enjoy!! 🧪

Graphical abstract of the paper which highlights The asymmetric state of bivalent nucleosomes regulates reader recruitment. Bivalency recruits H3K27me3 but not H3K4me3 readers, promoting a poised state. Bivalency also promotes recruitment of specific factors such as the acetylase KAT6B. KAT6B is required for proper activation of bivalent genes during differentiation

Super paper out of the Voigt group, that answers some longstanding questions about bivalent #histone marks!

They show asymmetric bivalency recruits repressive readers (but not H3K4me3 readers) and have emergent properties that neither of the two alone do (e.g. recruitment of KAT6B complex)!
🧪

Vertical Farms: From Vision to Reality Dr. Dickson Despommier believes vertical farming—the growing of crops indoors in multi-story urban buildings—can help feed the growing global population and undo the environmental damage caused by con...

Dickson was enormously optimistic and always sought to better the world around him. More than a microbiologist, he was an epidemiologist, environmentalist and a humanitarian working on some of the biggest challenges facing the world. Even writing a super interesting book about vertical farming:

Going back to the original TWiP and TWiV podcasts with Vincent Racaniello is such a treat! Highly recommend for anyone interested in non-diluted but approachable and fun microbiology:

www.microbe.tv/twip/twip-003/

www.microbe.tv/twiv/archive...

www.microbe.tv/twip/archive/