Interested in all things deep learning, biology, and molecular dynamics. @MSRNE #diversity & #representation matter (she/her)
Cambridge, MA
Joined April 2019
- Tweets1.4K
- Following1.1K
- Followers1.6K
- Likes8.7K
Pinned Tweet
I'm recruiting PhD interns for Summer 2027! If you’re excited about using generative protein language models to tackle real functional protein problems, please apply here!
Are you a PhD student interested in the intersection of machine learning and biology or healthcare?
Come do an internship with the BioML Group at MSR!
Make sure you submit a research statement!
apply.careers.microsoft.com/…
Sarah Alamdari retweeted
In new work, we report a surprising finding: prompting Evo 2 with a handful of sequences of shared function can steer the model to generate new sequences with similar function, a form of generative in-context learning (ICL). This is an emergent phenomenon that we did not anticipate during training.
In a first study, led by Alex Hao, we show that ICL enables few-shot design of mammalian regulatory elements, including novel enhancers with cell-type specificity for a neuronal subtype in the mouse brain.
In a second study, led by @garykbrixi and Alex Hao, we extend these results to the computational design of natural proteins, de novo antibodies, protein-RNA systems, and complete bacteriophage genomes.
As part of this work, we also release several useful new artifacts, including a highly efficient new inference stack. 1/n
Sarah Alamdari retweeted
Language models can learn new tasks from just a few examples in their prompt, with no retraining. This is called in-context learning (ICL). Now genome language models can do it too.
With a handful of examples, Evo 2 ICL generates proteins, de novo antibodies, CRISPR-Cas9 loci, mitochondrial genomes and entire bacteriophages.
We can even use the synthetic sequences iteratively to construct a chain-of-thought (CoT) and achieve harder design goals.
New work with Yukun A. Hao, @BrianHie and team 🧵
Sarah Alamdari retweeted
A model that generates an enzyme's reaction mechanism from its sequence, catalytic residues, substrates, and cofactors.
@arianemora_
Sarah Alamdari retweeted
Your molecule is really into your target protein. Unfortunately, it’s into other proteins too 💔
Introducing SelectBench: a benchmark for selectivity in generative structure-based drug design.
We started out thinking about algorithms for selectivity, then ran into a question: how do we evaluate them at scale on pharmacologically relevant examples? So we took a side quest and built a benchmark 🛠️ SelectBench is our first step toward evaluating in silico selectivity more systematically. Turns out, getting models to be picky is really hard!
Super proud of the entire team, especially Ella (@ERajaonson) and Yoon (@yololulu_) for leading the work. We can’t wait to present SelectBench in Sydney at NeurIPS 2026 as part of the Datasets and Benchmarks track. 🤗
Can your model design molecules with a type? 🧬
📄 Paper: arxiv.org/abs/2610.24124
💻 Code & benchmark: github.com/lovetaro/selectbe…
A molecule can bind the right protein and still be the wrong drug. Why? Because it might also bind proteins we don't want it to, potentially causing serious side effects.
📄 Paper: arxiv.org/abs/2610.24124
Sarah Alamdari retweeted
If you'd like to come work with me and my amazing colleagues as a full-time researcher @MSFTResearch @MSRNE, please apply here:
apply.careers.microsoft.com/…
apply.careers.microsoft.com/…
Sarah Alamdari retweeted
Great to see more groups are looking closer at microbial intergenic gold mine! Genomic context, again, is the critical information axis for discovery and interpretation.
Much of the functional biology encoded in genomes remains hidden in noncoding DNA.
In a preprint out today, @pdhsu and team introduce COMPASS, a genomic language model-based approach to systematically map this underexplored sequence space.
Sarah Alamdari retweeted
We’re hiring Senior and Principal researchers in ML & Stats at Microsoft Research New England @MSRNE! Amazing colleagues, remarkable breadth and depth of expertise, and unbeatable views. Come join us! Questions about the roles or the group? DM me!
Sarah Alamdari retweeted
1/n
Crossover episode! We introduce RFD4-Proteina, our latest design model - and the result of a year-long collaboration between the Baker Lab and NVIDIA GenAIR (the name says it all…). I hope you’ll agree - the results are astonishing. 🌐 Website: rfd4-proteina.org/
Kieran is having a generational protein design run this month 👀🔥
Avengers assemble: 1 year BakerLab + NVIDIA GenAIR = 🔥 rfd4-proteina.org
Presenting RFD4-Proteina, our latest protein design model, significantly better and faster than everything we compared against, both in silico and in the lab! 1/30
Sarah Alamdari retweeted
A step change in protein design: RFD4-Proteina is enables everyone today.
We partenered with @huggingface to make available a free space.
You can solve your hardest biology problem right now. No CS degree required: huggingface.co/blog/nvidia/r…
More to come...
Avengers assemble: 1 year BakerLab + NVIDIA GenAIR = 🔥 rfd4-proteina.org
Presenting RFD4-Proteina, our latest protein design model, significantly better and faster than everything we compared against, both in silico and in the lab! 1/30
Sarah Alamdari retweeted
If you're a PhD student interested in interning with me or one of my amazing colleagues at Microsoft Research New England (@MSRNE, @MSFTResearch) this summer, please apply here apply.careers.microsoft.com/…
Sarah Alamdari retweeted
Avengers assemble: 1 year BakerLab + NVIDIA GenAIR = 🔥 rfd4-proteina.org
Presenting RFD4-Proteina, our latest protein design model, significantly better and faster than everything we compared against, both in silico and in the lab! 1/30
Sarah Alamdari retweeted
Finally got GROMACS molecular dynamics simulations live-streaming directly into Blender using Molecular Nodes! Done via Docker, IMDv3
1.44 picoseconds/72 frames, all-atom molecular dynamics simulation of lysozyme in water, 303.15 K/30°C
#GROMACS #Blender3D #MolecularDynamics #MolecularNodes
During my PhD (not that long ago...) a single simulation on a single system could take weeks to months. It’s amazing to see that timeline collapsing so quickly!
Congrats to the team, looking forward to playing around with this!
Sarah Alamdari retweeted
Sharing a one-day workshop at the @MSRNE office on the frontier of AI for materials science during the @Materials_MRS fall meeting at Boston!
Scan the QR code to sign-up for poster and contributed presentations!
Sarah Alamdari retweeted
Public preview of BioEmu2! @MSFTResearch AI for Science’s new foundation model for biomolecular structure, dynamics and function. Open weights, MIT license.
github.com/microsoft/bioemu2
Movie: BioEmu2-simulated binding of intrinsically disordered peptide PUMA to MCL-1 receptor. 🧵⬇️
Deep sea gene editor is crazy
Sarah Alamdari retweeted
Continuing with with early life origins, Weiss et al. analyzed 6.1 million genes and found 355 protein families likely present in the last common universal ancestor, suggesting it was anaerobic, CO2-fixing, H2-dependent, N2-fixing, and thermophilic: nature.com/articles/nmicrobi…
Sarah Alamdari retweeted
🧬🌊 New preprint from our lab: DO1 Cas9, the first Cas9 nuclease characterised from the deep ocean — recovered from a metagenome at 1,500 m in the Atlantic.
❄️Cold-adapted: it cuts DNA efficiently at low temperature, where mesophilic SpCas9 retains only residual activity. 🧵1/4
Sarah Alamdari retweeted
Can we design proteins that cut other proteins? Our de novo cysteine protease paper is out in Nature! We designed new folds that cleave peptide backbones, with activity demonstrated in mammalian cells. Paper and code 👇
nature.com/articles/s41586-0…
