The Berger lab at @MIT_CSAIL works on a diverse set of problems in computational biology. Account run by lab members. Also @bergerlab.bsky.social.
MIT
Joined September 2020
- Tweets136
- Following25
- Followers2.4K
- Likes17
We are excited to present our new preprint "BatchRefiner: fast, significant improvement in batch integration of single-cell embeddings with ensemble refinement" by Daniel Schaffer, @kanghelenyihua, @ekinda, Daniel Edelman, and Bonnie Berger.
We tackle batch effects in scRNA-seq and scATAC-seq datasets by leveraging existing embedding methods and introduce centering, filtering, and scaling on each embedding dimension to improve batch correction and preserve biological signals.
BatchRefiner is shown to significantly improve batch integration across a wide range of embedding methods applied to scRNA-seq and scATAC-seq datasets.
preprint: biorxiv.org/content/10.64898…
Exciting work out of our lab!
@mihirbafna14 and I are excited to introduce Promera, a co-folding and design model with
• best-in-class binder filtering
• nanobody design with in-silico success rates matching hallucination
• case studies on hantavirus epitope targeting and GPCR agonism (1/8)
Excited to share some early steps from the lab to expand BindCraft-style protein design to structural state-switching! We think this could be an important step towards programmatically describing complex functions.
(led by @bjing2016 and @mihirbafna14 )
Binder design has come of age thanks to generative models—but how can we access the wider array of dynamic, multistate protein functions, so elegantly employed by nature?
@mihirbafna14 and I are excited to share SwitchCraft, a framework for designing such functions. (1/7)
Bonnie Berger Lab retweeted
Excited to share our paper on reformulating classical evolutionary dynamics when phenotypes are noisy! w/ Anna Sappington, @ShakhnovichLab, @lab_berger
Accepted to @RECOMBconf 2026
Featured today at @APSphysics DSNP Student Prize Session
Paper: biorxiv.org/content/10.64898…
🧵1/6
A recent profile on the 2024 chemistry Nobel laureates from our lab: pnas.org/doi/abs/10.1073/pna…. Check it out!
Bonnie Berger Lab retweeted
How can you evaluate agent trajectories with only black-box access to a verifier and the agent?
Introducing E-valuator: Reliable Agent Verifiers with Sequential Hypothesis Testing
❌no finetuning
❌no additional GPU compute
⬛black-box access
✔️controllable error rates w/ guarantees
Thread below:
Bonnie Berger Lab retweeted
What can you do when you need to evaluate a set of models but don't have too much labeled data? Solution: add unlabeled data. It was great to co-lead this project with @dmshanmugam and come talk to us at NeurIPS!
New #NeurIPS2025 paper: how should we evaluate machine learning models without a large, labeled dataset? We introduce Semi-Supervised Model Evaluation (SSME), which uses labeled and unlabeled data to estimate performance! We find SSME is far more accurate than standard methods.
We are thrilled to finally publish
@rohitsingh and @alexpywu's model of single-cell causality through the lens of a cell-state parallax! Now out in @NatureComms.
What does taking “portrait mode” photos have to do with decoding the genetic risk factors for Schizophrenia?
With GrID-Net, now out in @NatureComms, we show that they both benefit from using the parallax effect to reconstruct underlying “depth” from “flat” observations.
Excited to share our recent work: Sparse autoencoders uncover biologically interpretable features in protein language model representations now in PNAS. Thread below 🧵
8/ Paper: pnas.org/doi/10.1073/pnas.25…
MIT News article: news.mit.edu/2025/researcher…
Github: github.com/onkarsg10/Rep_SAE…
