- Vitrify: molecular simulations beyond DFT accuracy
- Biomni + Bayer: AI agent deployment for drug discovery
- Atomwell: Release of Bindwell's binding affinity model
The last 48 hours in #AIxBio
- Boltz + Bristol Myers Squibb: deploy biomolecular AI models
- Warp Sciences: AI x robotic CRO labs for drug discovery
- Carbon-X: Enables conversations with DNA, open model for DNA interpretation by hugging face
- Nanograb: AI for in vivo targeted drug delivery with lipid nanoparticles
- Holon-S1: Sophont and MedARC announce model for identifying cancer biomarkers from pathology slides
- Pinal: Model for protein design using plain language
This week in AIxBio
- Omnii-1 by Radical Numerics: AI-designed phage reaches human patient
- Claude designs scorpion toxin binders, yet to test in humans
- Baker Lab: RFD4-Proteina
- Chaperone: agent framework for protein design
- BioEmu2: protein simulations
How soon before we see the first cases of Drug Squatting?
Drug discovery is getting cheaper, its only a matter of time before random people start patenting drug designs instead of pursuing them..
Like Domain Squatting, drug squatting will soon become a pain for pharma..
@minchoi Very soon humans won't be able to distinguish between busllshit "discoveries" and real discovery, because there won't be anyone with enough knowlege to tell them apart..
@AnthropicAI Hey Anthropic, what category do research focused websites such as https://t.co/b67LhMSNT2 fall into? We use Claude for research on a regular basis?
@StanfordMed How is this different from uploading a paper on Claude and asking it questions related to the same?? Collaboration between two papers is not of any significance without the larger context..
AlphaFold: predicting the protein structure from amino acid sequence for >200 million proteins (awarded @NobelPrize)
Today
Alpha Genome Atlas: predicting the molecular impact of all 9 billion variants of the human genome
https://t.co/DkEtZofJ8s
Why Lab Automation is a nightmare?
Your $30 smart bulb speaks an open standard.
Your $300,000 liquid handler doesn't.
A thread on lab automation's most embarrassing failure
The vendors that treat openness as architecture โ open control, open state, open data, no API toll booth โ will own the AI-lab decade. The rest get designed out.
If your instrument needs its maker's permission to be useful, it's not an instrument. It's a subscription with a deck