Lightnews — Scholar-powered news

Leandros Boukas @leandrosboukas.bsky.social · Sep 8

One of the greats of 20th century biology. What a life.

www.youtube.com/watch?v=v9Em...

David Baltimore

YouTube video by InfiniteHistoryProject MIT

www.youtube.com

Leandros Boukas @leandrosboukas.bsky.social · Aug 8

yeah it's a very vanilla setting

Leandros Boukas @leandrosboukas.bsky.social · Aug 8

That's amusing for sure. But still, the approach is imo quite interesting and there's interesting results, e.g. the lower MAE for dispersion.

1

Leandros Boukas @leandrosboukas.bsky.social · Aug 7

This looks like a creative use of transformers. Looking forward to diving into this.
arxiv.org/abs/2508.04111

Negative binomial regression and inference using a pre-trained transformer

Negative binomial regression is essential for analyzing over-dispersed count data in in comparative studies, but parameter estimation becomes computationally challenging in large screens requiring mil...

arxiv.org

1 2 8

Reposted by Leandros Boukas

John Greally @greally.bsky.social · Aug 2

My book Epigenetics: History, Molecules and Diseases will be published in exactly one month (September 2).

Finalizing cover (looks great), probably going to printer this week.

Nervous anticipation is the mood right now.

6 9 76

Reposted by Leandros Boukas

Annemieke Aartsma-Rus @oligogirl.bsky.social · May 27

#apaperaday is back from holiday. Yuzu is still with the birbsitter so a holiday picture it is. Today's pick is from @nejm.org by Musunuru et al on an N=1 case (patient with severe metabolic disease) treated by base editing (CRISPR therapy). DOI: 10.1056/NEJMoa2504747 Very good way to restart!

1 3 8

Reposted by Leandros Boukas

Eric Topol @erictopol.bsky.social · May 18

This week's genome editing triumph is a big deal.
Here's why
erictopol.substack.com/p/the-first-...

Multiple headlines about the first CRISPR 2.0 personalized genome editing and photo of the baby treated

23 140 630

Reposted by Leandros Boukas

Richard Sever @richardsever.bsky.social · May 18

Reminded of Ed Yong’s response to this question:

“If I cover a preprint, I talk to 2-3 experts to get their views before writing the story.

That’s exactly what I do if I cover a journal article…”

Nicholas Bauer PhD @bioturbonick.net · May 18

Absolutely.

Preprints are for communicating among scientists.

If a journalist wants to cover one, they should commission a peer review.

Chemjobber @chemjobber.bsky.social · May 18

Resolved: preprints should not be covered in the news, no matter how tantalizing the results are

6 31 96

Leandros Boukas @leandrosboukas.bsky.social · May 16

Looking forward to this next week

FragileNucleosome @fnucleosome.bsky.social · May 15

Join us on Wednesday next week for two exciting talks on transcription regulation from @kasitc.bsky.social and @davidsuter.bsky.social!

You can register at: us06web.zoom.us/webinar/regi...

2

Leandros Boukas @leandrosboukas.bsky.social · May 9

Beautiful combination of population-scale genetic analyses and mouse work. Damaging variants in a highly constrained chromatin remodeler (CHD1) are better tolerated in males, with mouse experiments indicating a protective effect of androgens.

medRxivpreprint @medrxivpreprint.bsky.social · May 8

Androgens mediate sexual dimorphism in Pilarowski-Bjornsson Syndrome https://www.medrxiv.org/content/10.1101/2025.05.06.25326635v1

Reposted by Leandros Boukas

Lars Velten @larsplus.bsky.social · May 8

Out in Cell @cp-cell.bsky.social: Design principles of cell-state-specific enhancers in hematopoiesis
🧬🩸 screen of fully synthetic enhancers in blood progenitors
🤖 AI that creates new cell state specific enhancers
🔍 negative synergies between TFs lead to specificity!
www.cell.com/cell/fulltex...
🧵

Design principles of cell-state-specific enhancers in hematopoiesis

Screen of minimalistic enhancers in blood progenitor cells demonstrates widespread dual activator-repressor function of transcription factors (TFs) and enables the model-guided design of cell-state-sp...

www.cell.com

4 57 140

Reposted by Leandros Boukas

Jeff Leek @jtleek.bsky.social · Apr 30

Check this out - the company I co-founded is trying to figure out how to make this super easy product.synthesize.bio (and we have beta models out to AI generate new expression data based on experimental design!)

Bioinformatics Platform | NGS Data Analysis - Synthesize BioArrow Right

Experience a next-gen analysis tool tailored for biology · Enhance discovery with our dynamic visualization canvas · A central space for bioinformatics teamwork.

product.synthesize.bio

4 7

Reposted by Leandros Boukas

Michael Love @mikelove.bsky.social · Apr 8

+1 Anshul

Part of this means reviewers not ripping on manuscripts that explicitly mention limitations *for those limitations*

4 24

Reposted by Leandros Boukas

Anshul Kundaje @anshulkundaje.bsky.social · Apr 8

Just a gentle reminder that deceptive hyping in scientific publications (which includes preprints) is actually antithetical to the core mission of the scientific process. We can stay grounded, truthful, humble while being ambitious. Revealing caveats, pitfalls & limitations speeds up progress.

3 16 95

Reposted by Leandros Boukas

Gang Fang @gangfang.bsky.social · Mar 30

Great new review @natrevgenet.bsky.social by @sedlazeck.bsky.social @timp0.bsky.social and @yileifu.bsky.social! 👇highlighting both the promise of long read DNA methylation analysis and remaining challenges such as the difficulty to benchmark 5hmC and non-CG 5mC given their low abundance👏

Yilei Fu @yileifu.bsky.social · Mar 29

We summarized state-of-the-art computational methods for DNA methylation analysis using long-read sequencing, covering everything from base calling to sample-level, cell-type-level, and even population-scale analysis. Huge thanks to @sedlazeck.bsky.social and @timp0.bsky.social for this great work!

Nature Reviews Genetics @natrevgenet.nature.com · Mar 28

New online! Computational analysis of DNA methylation from long-read sequencing

6 14

Leandros Boukas @leandrosboukas.bsky.social · Mar 24

This is really a fantastic piece on causality in biological systems, and relevant beyond neuroscience
onlinelibrary.wiley.com/doi/10.1111/...

Beyond Mechanism—Extending Our Concepts of Causation in Neuroscience

The search for neural mechanisms of behaviour often relies on a synchronic, driving view of causation, where neural activity drives more neural activity, which eventually drives behaviour. The real c...

onlinelibrary.wiley.com

1 1

Reposted by Leandros Boukas

Stefan Schoenfelder @stefanschoenfelder.bsky.social · Mar 21

Plenty of work required to rigorously test the proposed hypothesis but my initial thoughts are this perspective by Adrian Bird in @cp-molcell.bsky.social may well be the most significant conceptual advance in the 3D genome organisation field in years:
www.sciencedirect.com/science/arti...

Cohesin as an essential disruptor of chromosome organization

Cohesin is a multi-subunit molecular machine that is able to create lateral chromatin loops within a linear chromosome fiber. Despite intense study, a…

www.sciencedirect.com

3 9 29

Reposted by Leandros Boukas

Tuuli Lappalainen @tuuliel.bsky.social · Mar 19

I just the word epigenetics; it has so many meanings that it's next to useless. I prefer less ambigous terms: Epigenomic data for chromatin-related data (including methylation); gene regulatory mechanisms for the general phenomenon of how genes are regulated, etc.

3 4 7

Reposted by Leandros Boukas

Nikolai Slavov @slavov-n.bsky.social · Mar 1

Remember the role of regulation by degradation.

Research often focusses on the regulation of synthesis rates.

Here, @andrewleduc.bsky.social and I show that degradation can be just as influential in shaping protein abundance variation.

www.biorxiv.org/content/10.1...

Protein degradation and growth dependent dilution substantially shape mammalian proteomes

Cellular protein concentrations are maintained through a balance of synthesis and clearance. Clearance occurs through both protein degradation and growth-dependent dilution. At slow growth, clearance ...

www.biorxiv.org

6 16

Reposted by Leandros Boukas

Wolfgang Huber @wkhuber.bsky.social · Feb 27

The analogy of omics data normalization and cooking:
some processing is usually necessary, overcooking makes it bland, highly over-processed foods are unhealthy, and the quality of the ingredients matters.

1 23 76

Leandros Boukas @leandrosboukas.bsky.social · Feb 27

Important to mention that duoNovo was developed and tested using sequencing data from the NIH-funded @gregor-research.bsky.social consortium

Feedback and comments are especially welcome - as I mentioned earlier, we really hope that duoNovo can be broadly useful (n/n)

1

Leandros Boukas @leandrosboukas.bsky.social · Feb 27

In the preprint we show that this simple approach performs very well in practice. Among rare variants - which are of course the most likely to be disease-causing - duoNovo has perfect accuracy in our evaluations (7/n)

1

Leandros Boukas @leandrosboukas.bsky.social · Feb 27

Then, using just one parent, we can tell whether the haplotype containing a given variant of interest was inherited from the available parent. If that's the case, then the variant must be de novo - we don't need to look at the other parent. (6/n)

1

Leandros Boukas @leandrosboukas.bsky.social · Feb 27

Our method solves this problem using a simple idea. We first reconstruct haplotypes using read-backed phasing enabled by long read sequencing. This allows us to phase variants that we want to test for de novo status (which isn't possible with non-read-backed phasing approaches). (5/n)

1

Leandros Boukas @leandrosboukas.bsky.social · Feb 27

Which is why de novo status is used as a strong criterion of variant pathogenicity.

However, to identify a variant as de novo, both biological parents have to be sequenced. For millions of families, this is impossible for a variety of reasons. (4/n)

1