Ron Alfa

Announcing OCTO-VirtualCell (vc) a multi-scale, multimodal transformer trained to predict gene expression for a virtual cell in cellular contexts within patient tissue samples. Complete wth the Celleporter demo app to explore the data! 1/

The level of enthusiasm for our foundation models from teams that spend their days in the trenches of translational medicine is gratifying.

Abstract deadline for CSH Symposium on AI in Biology is next week, Mar 26! This frontier AI x Bio meeting will span many scales of biology, from proteins and genomes to cells, medicine, and agents, with 50+ invited speakers who are leaders in the field. meetings.cshl.edu/abstracts.aspx…

This is how simulations with @NOETIK_ai foundation models get to new biological discoveries, from patient data.

Correct

It's hard to even imagine the true scale of impact for foundation models of human biology - transformative. And there are indicators we are on the timeline.

@iScienceLuvr @BoWang87 @NVIDIAGTC @Prof_Lundberg @DBBurkhardt Had to head back to office unfortunately.

@BoWang87 @NVIDIAGTC @Prof_Lundberg @Ronalfa @DBBurkhardt Are you still around in San Jose today or tomorrow? Would love to meet!

Had a fun + packed panel this morning at @NVIDIAGTC with @Prof_Lundberg, @Ronalfa, and Phil Lorenz. Hot takes flying on scaling laws in biology — but the love for 🦞 seems universal 😅 Huge shoutout to @DBBurkhardt for the fantastic moderation 🙌 Hopefully the recording will be made public soon.

@BoWang87 @NVIDIAGTC @Prof_Lundberg @DBBurkhardt So fun! @DBBurkhardt crushed it.

@BoWang87 @NVIDIAHealth @NVIDIAAI 💯

@Ronalfa @NVIDIAHealth @NVIDIAAI It should be fun! Look forward

Catch my panel with @BoWang87, Emma Lundberg, Philipp Lorenz, Daniel Burkhardt on bio x ai and scaling laws this morning at 10a at GTC. @NVIDIAHealth @NVIDIAAI

100%. Working hard on this problem every day.

@0xfrankly We have a virtual cell world model but all our data are patient tissues and we are entirely focused on simulating biology from spatial contexts. We don’t do any in vitro biology.

@Ronalfa Is Noetik building a virtual cell as well?

New in vitro perturbation virtual cell model from Xaira, with scaling. Others in this space: Recursion, Tahoe, Genentech, Arc, BioHub

LinkedIn is like @moltbook except the bots have humans post for them.

This is a ton of Blackwells, impressive. Indicator of massive investments pharma will be making AI.

@hoifungpoon Appreciate your work Hoifung. Congrats on the great pub in Cell. Let’s train a massive model on thousands of patients!

@Ronalfa As someone already pointed out in the replies, our training data comprises 40 million cells generated on whole slides, not TMAs. The GigaTIME paper has been published in Cell and the GigaTIME model is open-weight.

Wow this is trained on 21 patients data in 1 cancer. 😬 For context we are training these types of models on almost 4,000 patients across all modalities paired. “Our training data comprises of data collected from 21 patients across different stages of lung adenocarcinoma.”

@ClarkPolner I think you just need enough quality data and modalities to generate strong representations of biology. I'm sure there are different paths to this. Also large diversity of tissue is important.

That much I knew; was thinking more about all the contextual data around the diseased tissue // how much (and profiled in what modalities and dimensions) do you think is necessary to get to the pt where you have a world model that gets you that unlock (or does disease alone do it?)

Massive unlock for biology comes from LLM-powered agents reasoning and running virtual experiments with human bio world models.

@ClarkPolner We generate the pretraining data from human tissues.

@Ronalfa From where do the data come for the human bio world models

Reasons to be pessimistic (and optimistic) on the future of biosecurity owlposting.com/p/reasons-to-b… "It was such a fun read (if you can say that about an article on weapons)!" —a glowing review from an early reader this is (once again) the longest article I have ever published at 13,000 words. it involves interviews with 16+ researchers/VC's/policy folks in this field, and discusses basically every single facet of biosecurity that i could find. topics include: how machine-learning in rapid response therapeutic design may work, the financial status of the customer base of biosecurity startups, why agroterrorism feels extremely likely to me, and a lot more i admittedly started the essay pessimistic that this subject matters at all, and i end it surprised that it doesn't keep more people awake at night. im not a doomer about it all, but i can see how people become one. very grateful to the people who decide to spend their career (or some fraction of it) working here, and especially grateful to the ones who helped teach me about the subject