Lightnews — Scholar-powered news

Jakub Bartoszewicz @jmbartoszewicz.bsky.social · Jul 23

Photo credit by the one and only @melanianowicka.bsky.social. Visit her presenting our collab work on risk prediction for AI in synthetic biology: poster C-411!

1

Jakub Bartoszewicz @jmbartoszewicz.bsky.social · Jul 23

Having a blast at #ismbeccb2025! Let's catch up if you're into AI for SynBio, bioLMs, phages, microbes, biosecurity. We're also always looking for PhD students/postdocs! Case in point: multiple 3yr postdoc positions. Work with us or other PIs at HPI near Berlin 🇪🇺!
lnkd.in/eD-ueQmV
lnkd.in/eASJEZAW

2 2

Reposted by Jakub Bartoszewicz

European Synthetic Biology Society @eusynbios.org · Mar 10

🧬 Passionate about #SyntheticBiology and eager to make an impact? There's still time to get involved and join us!✨

⏳Application deadline extended until March 14th!

👇 Open positions & application info:

European Synthetic Biology Society @eusynbios.org · Feb 24

🚀 Join the #EUSynBioS Steering Committee!

🧬 Passionate about #SyntheticBiology? Based in #Europe? Help grow our community!

We are opening new positions, check them out & apply here 👉 linkedin.com/feed/update/...

#SynBio #ScienceCommunication #Policy

European Synthetic Biology Society (EUSynBioS) on LinkedIn: #syntheticbiology #europe #synbio #europe #outreach #communication…

🧬 Are you passionate about #SyntheticBiology, based in #Europe and willing to help our growing community? ⭐️ 𝗝𝗼𝗶𝗻 𝘁𝗵𝗲 𝗘𝗨𝗦𝘆𝗻𝗕𝗶𝗼𝗦…

linkedin.com

7 3

Reposted by Jakub Bartoszewicz

Hannes Stark @hannes-stark.bsky.social · Mar 10

New paper (and #ICLR2025 Oral :)):
ProtComposer: Compositional Protein Structure Generation with 3D Ellipsoids arxiv.org/abs/2503.05025

Condition on your 3D layout (of ellipsoids) to generate proteins like this or to get better designability/diversity/novelty tradeoffs.
1/6

1 5 18

Reposted by Jakub Bartoszewicz

Mike White @genologos.bsky.social · Mar 4

Can generative AI write functional genomes? We don't know, because what's missing in this field is a thorough program of experimental testing.

thisgenomiclife.substack.com/p/can-genera...

Can generative AI write genomes on demand?

A reality check for generative DNA models

thisgenomiclife.substack.com

2 4 13

Jakub Bartoszewicz @jmbartoszewicz.bsky.social · Feb 26

Want to push the frontier on AI for safe synthetic biology and infectious disease? Just extended the deadline for PhD/Postdoc positions: apply by March 3rd!

Jakub Bartoszewicz @jmbartoszewicz.bsky.social · Feb 11

I'm setting up my lab at @hpi.bsky.social, on the edge of vibrant Berlin! PhD/postdoc positions in AI for safe synthetic biology & infectious diseases (AMR), in collab with @melanianowicka.bsky.social.

Apply by Feb 24: email [email protected] with “[Open-25]” in the subject. DM for details!

Reposted by Jakub Bartoszewicz

Jean-Paul Pirnay @jeanpaulpirnay.bsky.social · Feb 17

Check out our latest #phagetherapy review in Nature Reviews Methods Primers.

rdcu.be/d9H92

Happy to share a conceptual schematic for a “phrinter (phage printer)” 😊

Many thanks to all co-authors, editors and reviewers!

1 6 8

Reposted by Jakub Bartoszewicz

Peter Koo @pkoo562.bsky.social · Feb 5

[SAVE THE DATE] MLCB 2025 is happening Sept 10-11 at the NY Genome Center in NYC!

Attend the premier conference at the intersection of ML & Bio, share your research and make lasting connections!

Submission deadline: June 1
More details: mlcb.github.io

Help spread the word—please RT! #MLCB2025

1 27 41

Jakub Bartoszewicz @jmbartoszewicz.bsky.social · Feb 11

Sounds good? Apply with your CV, transcripts, contact info for 2 references & a brief statement of interests/motivation. Join us to enjoy your research in one of the most vibrant cities on the planet, with a healthy work-life balance and many opportunities for impact and growth!

1

Jakub Bartoszewicz @jmbartoszewicz.bsky.social · Feb 11

We're especially interested in generative models & LMs for bio, biosecurity, interpretability, & good OOD generalization (applied to hard biological tasks). The positions come with a generous tax-free scholarship. Of course, candidates from all backgrounds are welcome - please share!

1 1

Jakub Bartoszewicz @jmbartoszewicz.bsky.social · Feb 11

I'm setting up my lab at @hpi.bsky.social, on the edge of vibrant Berlin! PhD/postdoc positions in AI for safe synthetic biology & infectious diseases (AMR), in collab with @melanianowicka.bsky.social.

Apply by Feb 24: email [email protected] with “[Open-25]” in the subject. DM for details!

2 5 12

Reposted by Jakub Bartoszewicz

Kevin K. Yang 楊凱筌 @kevinkaichuang.bsky.social · Feb 10

A protein language model trained to predict subcellular localization for human proteins can generate de novo sequences with the desired localization and identify pathological mutations.

@itamarchinn.bsky.social @pgmikhael.bsky.social

www.science.org/doi/10.1126/...

ProtGPS classifies protein compartment with high performance.

Generative modeling creates novel proteins that concentrate in a desired condensate.

Pathogenic mutations are predicted to alter protein compartmentalization

2 18 93

Reposted by Jakub Bartoszewicz

Mohammed AlQuraishi @moalquraishi.bsky.social · Feb 10

It's long seemed that molecular biology is a natural home for ML interpretability research, given the maturity of human-constructed models of biological mechanisms—permitting direct comparison with their ML-derived counterparts—unlike vision and NLP. Our first foray below👇.

Etowah Adams @etowah0.bsky.social · Feb 10

Can we learn protein biology from a language model?

In new work led by @liambai.bsky.social and me, we explore how sparse autoencoders can help us understand biology—going from mechanistic interpretability to mechanistic biology.

15 38

Reposted by Jakub Bartoszewicz

Diego del Alamo @delalamo.xyz · Feb 10

This might be the best paper on applying sparse autoencoders to protein language models. The authors identify how neural networks trained on amino acid sequences "discover" different features, some specific to individual protein families, other for substructures

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

A range of features identified from sparse autoencoders trained on different layers on ESM2-650M

2 11 44

Reposted by Jakub Bartoszewicz

Yunha Hwang @microyunha.bsky.social · Dec 17

Can LLM agents discover novel protein functions? Introducing Gaia Agent 🌎 🤖: an AI biologist capable of reasoning across genomic contexts to predict functions of proteins! Gaia Agent is now integrated with Gaia Search at gaia.tatta.bio

2 13 38

Reposted by Jakub Bartoszewicz

Vaughn Cooper @vscooper.micropopbio.org · Dec 13

In which esteemed colleague Jeremy Berg discusses how his contributions to racemic protein synthesis, inspired by Pauling, demonstrated that D-proteins are immunologically invisible #mirrorbiology

Jeremy Berg @jeremymberg.bsky.social · Dec 11

Pauling’s chapter ends “In ‘Through the Looking Glass' Alice said 'Perhaps looking-glass milk isn’t good to drink…Nobody knew that proteins are built of the left-handed amino acids; but Alice was justified in raising the question. The answer is that looking -glass milk is not good to drink.'

25/n

6 10

Reposted by Jakub Bartoszewicz

Tom Inglesby @t-inglesby.bsky.social · Dec 12

Sharing here a new commentary in Science Policy Forum that provides new analysis on risks related to the potential future creation of mirror bacteria — synthetic organisms in which all molecules have reversed chirality (i.e. are ‘mirrored’). 1/x

www.science.org/doi/10.1126/...

Confronting risks of mirror life

Broad discussion is needed to chart a path forward.

www.science.org

3 18 29

Reposted by Jakub Bartoszewicz

Jian Ma @jianma.bsky.social · Dec 11

Can we bypass the resource bottleneck of pretraining genomic Foundation Models? Our work L2G repurposes language LLMs for genomics via cross-modal transfer, matching fine-tuned genomic FMs. Kudos to Wenduo & fantastic collab w/ @atalwalkar.bsky.social. L2G, language to genome; L2G, life’s too good!

bioRxivpreprint @biorxivpreprint.bsky.social · Dec 11

L2G: Repurposing Language Models for Genomics Tasks https://www.biorxiv.org/content/10.1101/2024.12.09.627422v1

3 9

Reposted by Jakub Bartoszewicz

Anshul Kundaje @anshulkundaje.bsky.social · Dec 11

Check out this systematic benchmark of genome-wide, annotation agnostic DNALMs & strong baseline ab-initio models for biologically meaningful tasks in regulatory genomics 1/

Austin Wang @austintwang.bsky.social · Dec 11

(1/10) Excited to announce our latest work! @arpita-s.bsky.social, @amanpatel100.bsky.social , and I will be presenting DART-Eval, a rigorous suite of evals for DNA Language Models on transcriptional regulatory DNA at #NeurIPS2024. Check it out! arxiv.org/abs/2412.05430

DART-Eval: A Comprehensive DNA Language Model Evaluation Benchmark on Regulatory DNA

Recent advances in self-supervised models for natural language, vision, and protein sequences have inspired the development of large genomic DNA language models (DNALMs). These models aim to learn gen...

arxiv.org

5 23 61