Sabitlenmiş Tweet
Excited to share my first contribution here at Illumina! We developed PromoterAI, a deep neural network that accurately identifies non-coding promoter variants that disrupt gene expression.🧵 (1/)
English
Gherman Novakovsky (слава Україні! 🇺🇦)
772 posts
Gherman Novakovsky (слава Україні! 🇺🇦)
@NovakovskyG
PhD, Illumina AI lab; interested in Deep Learning and genome regulation; also drawing, martial arts, guitar, and death metal! (he/him)
Katılım Ocak 2018
398 Takip Edilen273 Takipçiler
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
This administration entire policy is to torture hard working people who actually contribute to the nation. This will lead to faster decline & push even more skilled immigrants to other nations. A special congrats to tech/biotechMAGA.
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Rare disease diagnoses can rely on exome sequencing, but answers may be hiding in noncoding regions. 🧬 PromoterAI is a new deep learning tool that identifies pathogenic promoter variants, which may account for up to 6% of rare disease genetic burden 🔍
science.org/doi/10.1126/sc…
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
In a saturation MPRA of the MAPT promoter, PromoterAI tracked measured variant effects, supporting its use for prioritizing pathogenic promoter variants.
biorxiv.org/content/10.648…
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
When linear models and deep learning give the same results, orders of magnitude more training data are needed nature.com/articles/s4159… ;
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Excited to share our RegVelo paper in Cell
cell.com/cell/fulltext/…
We unify RNA velocity + GRNs into one model → better OOD prediction of perturbations (e.g. gene KOs), with examples incl. neural crest KO predictions 🔬
Big thanks to W Wang, Z Hu & T Sauka-Spengler 🙏
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Exciting new insights on CpG islands (CGIs) regulation by transcription factors (TFs)! CGIs drive most transcription initiation with unclear regulation. We find that chromatin-opening TFs are key players—following a surprisingly simple rule.
biorxiv.org/content/10.648…
1/9
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
1/ PromoterAI scores are now viewable as a track in @GenomeBrowser. Below is a ClinVar example that clearly highlights the value:
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Our Human Multiomic Development Atlas paper is out in Nature today! A heart-felt "thank you" to all co-authors for their tireless work on this complex yet exciting project! Congrats all! nature.com/articles/s4158…
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
It’s well known that inflammation increases cancer risk, but how?
The answer: the epigenome "remembers" inflammation and primes stem cells for cancer.
Here is our paper: nature.com/articles/s4158…
And a special shoutout to the lead author @snaga13
A 🧵
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
We are also releasing self-contained lecture notes that explain flow matching and diffusion models from scratch. This goes from "zero" to the state-of-the-art in modern Generative AI.
📖 Read the notes here: arxiv.org/abs/2506.02070
Joint work with @EErives40101.
Peter Holderrieth@peholderrieth
🚀MIT Flow Matching and Diffusion Lecture 2026 Released (diffusion.csail.mit.edu)! We just released our new MIT 2026 course on flow matching and diffusion models! We teach the full stack of modern AI image, video, protein generators - theory and practice. We include: 📺 Videos: Step-by-step derivations. 📝 Notes: Mathematically self-contained lecture notes 💻 Coding: Hands-on exercises for every component We fully improved last years’ iteration and added new topics: latent spaces, diffusion transformers, building language models with discrete diffusion models. Everything is available here: diffusion.csail.mit.edu A huge thanks to Tommi Jaakkola for his support in making this class possible and Ashay Athalye (MIT SOUL) for the incredible production! Was fun to do this with @RShprints! #MachineLearning #GenerativeAI #MIT #DiffusionModels #AI
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Models are typically specialized to new domains by finetuning on small, high-quality datasets.
We find that repeating the same dataset 10–50× starting from pretraining leads to substantially better downstream performance, in some cases outperforming larger models. 🧵
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Great to the see the flurry of single gene knockdown Perturb-seq like atlases from cell-lines, mouse brain etc over the last few days. These are undoubtedly very valuable datasets. I just want to re-iterate a few other very important expt. design considerations 1/
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
@SashaGusevPosts @javier_maravall Gaps *this* wide have been shown before, in Figure 2D, for splice variant effect prediction (SpliceAI is 700K parameters). The x-axis ranges from 0 to 1 here so it may not be immediately apparent, but its the same 0.6 to 0.9 gap.
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Nature research paper: Functional dissection of complex trait variants at single-nucleotide resolution
go.nature.com/4ck0zkd
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Can we simulate realistic evolutionary trajectories and “replay the tape of life”? In this work, we propose a flexible, generalizable framework for modeling how the entire protein seq evolves over time while capturing complex interactions across sites. 1/n
doi.org/10.64898/2026.…
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Nature research paper: Regulatory grammar in human promoters uncovered by MPRA-based deep learning
go.nature.com/3NUvNEh
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
AlphaGenome is out in @nature today along with model weights! 🧬
📄 Paper: nature.com/articles/s4158…
💻 Weights: github.com/google-deepmin…
Getting here wasn’t a straight path. We sat down @googledeepmind to discuss the story behind the model, paper & API: youtu.be/V8lhUqKqzUc
YouTube
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
Molecular Genetics @ University of Toronto 🇨🇦@UofT🦠 is recruiting an Assistant Professor in #Virology!!
Come build your dream lab!
🔗jobs.utoronto.ca/job/Toronto-As…
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
A paper in Nature presents a detailed map of human chromosomes within the nucleus. This resource provides a foundation for an improved understanding of how the physical layout of human DNA is associated with biological expression. go.nature.com/44uMmft
English
Gherman Novakovsky (слава Україні! 🇺🇦) retweetledi
I've been thinking about the "virtual cell" concept and wanted to write up a few thoughts. Specifically on how I think the prior experience in GWAS informs the most likely way these models will be useful.
andrewcarroll.github.io/2025/12/23/the…
English