Keep pulling the thread on Patrick.
The ARC Institute's primary goal is to create "virtual cells" and simulate human biology using foundation models.
The ultimate goal of the ARC Institute's virtual cell project is to enable in silico target identification for discovering new drug targets and their corresponding drug compositions.
A key benchmark for a "GPT-3 moment" in virtual cell models would be the ability to predict that the four Yamanaka factors can reprogram a fibroblast into a stem-like state, effectively rediscovering a Nobel Prize-winning finding.
The market capitalization gained by Eli Lilly and Novo Nordisk from GLP-1 drugs, estimated at over a trillion dollars, exceeds the combined market cap of all biotech companies founded in the last 40 years.
Dario Amadei, in his essay "Machines of Love and Grace," predicted that AI could lead to the prevention of many infectious diseases and a doubling of human lifespans within the next decade.
Patrick suggests that based on an observed eight-year cycle of major architectural shifts in deep learning, a new fundamental architecture is overdue and could emerge around 2025.
The ARC Institute was founded as an organizational experiment to co-locate researchers from neuroscience, immunology, machine learning, chemical biology, and genomics to foster interdisciplinary collaboration.
The ARC Institute has two flagship projects: one focused on identifying drug targets for Alzheimer's disease and another dedicated to creating virtual cells.
The AlphaFold algorithm can predict the final 3D structure of a protein from its amino acid sequence with over 90% accuracy.
Patrick assesses the current state of virtual cell modeling capabilities as being analogous to the GPT-1 to GPT-2 stage of large language model development.
Approximately 90% of drugs fail during clinical trials.
There is a growing trend for pharmaceutical companies to conduct Phase 1 clinical trials overseas and then use that data to support Phase 2 efficacy trials in the United States.