Pinned Tweet
Today I’m excited to present antiGen, a framework for anticipating future evolutionary changes to viral proteins by learning from their evolutionary past. Key applications include vaccine and therapeutic design.
Paper: biorxiv.org/content/10.64898…
Ivan Specht retweeted
1/ AI agents can now accomplish many research tasks. But tackling frontier problems requires deciding which questions matter and when to trust results.
We're excited to introduce DiscoveryAgent to empower scientists in their pursuit of rigorous, open-ended discovery 🧵
Ivan Specht retweeted
Excited to share Germinal Multi, a framework for multi-target antibody design and rescue. We applied it to cancer resistance mutations, historical SARS-CoV-2 evolution, and paired with a viral evolutionary model, antiGen, to design against a forecasted escape variant! (1/8)
antiGen predictions can serve as the basis for antibody design tasks, as shown by Soyoon Park, @talaldotpdb, and colleagues: biorxiv.org/content/10.64898….
1
4
137
Grateful to have been part of such a wonderful team: Soyoon Park, @aakarshv1, @SeyoneC, @driscoll_cl, @garykbrixi, Julia Palacios, and @BrianHie. Check out our blog post: arcinstitute.org/news/antige….
5
145
Ivan Specht retweeted
We present JUNIPER, our outbreak reconstruction tool that incorporates within-host variants, models missing data, and scales to large, sparsely sampled datasets to achieve state-of-the-art performance. Led by @ivan_specht et al. @sabeti_lab. medrxiv.org/content/10.1101/… 1/12 🧵
Ivan Specht retweeted
(1/4) Excited to introduce ProCyon: a multimodal foundation model to model, generate, and predict protein phenotypes, led by stellar @oq_35 @YepHuang Robert Calef @valegiunca
👉 biorxiv.org/content/10.1101/…
ProCyon is an 11B parameter multimodal model that integrates protein sequences, structures, and natural language
ProCyon models, predicts, and generates phenotypes with zero-shot task transfer, breaking past the limits of fixed vocabularies or ontologies
🔬 The Challenge: Predicting protein structure is advanced, but predicting protein phenotypes—observable traits linking molecular function to biology—is still an open problem. 40% of human proteins lack context-specific insights, while 20% remain almost entirely uncharacterized
Ivan Specht retweeted
Talent shines in the Sabeti Lab! ✨ Ivan Specht's senior thesis, 'Reconstructing Viral Epidemics: A Random Tree Approach,' is now featured on the Harvard Math website. Dive into Ivan's groundbreaking work and join us in celebrating his achievement! 🎉
rb.gy/zvfscu
Ivan Specht retweeted
Meet @ivan_specht! Ivan's work in the @sabeti_lab involves epidemiological modeling of infectious diseases as a means to inform public health policy. Read more about his work here: buff.ly/3sTACmW #OperationOutbreak #STEM #education
Thrilled to be here at #NCTMDC23 sharing Operation Outbreak (@OutbreakEdu) with math teachers from all around the country! Operation Outbreak is an outbreak simulation technology that spreads a ‘virtual pathogen’ between smartphones via Bluetooth to teach epidemiology.
10
578
We developed a new model for within-host viral evolution, which we used to reconstruct outbreak transmission networks. A thread on our genomic epidemiology paper with @PetrosBrittany, @GageKMoreno et al. @sabeti_lab medrxiv.org/content/10.1101/…
Thanks to our authors, full list: @ivan_specht, @PetrosBrittany, @GageKMoreno, @tbrockfi, @lakrasil, @mschifferli, Katherine Yang, Paul Cronan, @orglennon, …
1
1
273
…@glipsnort, @DannyJPark, @bronwynmacinnis, Al Ozonoff, @ben_fry, Michael Mitzenmacher, Patrick Varilly, and @PardisSabeti.
1
272










