One-liner
Funding the open source development of a multimodal Virtual Cell foundation model to accelerate longevity research and drug discovery by enabling predictive, large-scale cellular perturbation simulations.
Team
Principal Investigator: AbuGoot Lab, VitaDAO, Prime Intellect
Simple Summary
TL;DR
We propose funding the development of the Virtual Cell AI foundational model that integrates multi-modal cellular data, including single-cell RNA-seq, ATAC-seq, and protein data, to predict cellular responses to perturbations. This will enhance our ability to model disease mechanisms, optimize drug discovery, and simulate experimental conditions currently infeasible at scale. Scientists in VitaDAOs ecosystem will be connected to the AbuGoot Lab to drive longevity research discovery utilizing the open source foundation model.
Summary
Biological research is rapidly evolving with AI-driven predictive modeling. However, existing cellular models struggle with multimodal integration and scalability. The AbuGoot Lab, in collaboration with VitaDAO and Prime Intellect, aims to develop a high-resolution Virtual Cell foundation model trained on large-scale single-cell and multimodal datasets. This model will:
- Predict drug efficacy and toxicity in aging-related diseases.
- Simulate disease states and interventions, reducing the need for costly wet-lab experiments.
- Generate open-access in silico perturbation datasets for the scientific community.
This initiative will empower researchers to explore cellular regulation, gene expression, and drug interactions at an unprecedented scale, ultimately accelerating the development of longevity therapeutics.
Problem
Longevity and aging research require extensive experimentation to identify therapeutic targets. Traditional wet-lab methods are slow, expensive, and limited in scalability. Despite advances in single-cell technologies, foundational models for predicting cellular responses remain inadequate due to:
- Data heterogeneity – Integrating multimodal datasets remains a challenge.
- Scalability issues – Existing models fail to generalize across diverse cell types and perturbations.
- Limited causal inference – Most models lack robust predictive power for novel interventions.
Without a robust predictive framework, progress in longevity drug discovery remains constrained.
Solution
Our Virtual Cell model will address these challenges by:
- Developing a foundational model trained on large-scale single-cell and multimodal datasets (RNA-seq, ATAC-seq, protein expression, etc.).
- Leveraging causal AI frameworks to predict perturbation responses across healthy and diseased cell states.
- Creating an open-access database of in silico perturbation effects, democratizing access to cutting-edge predictions for researchers globally.
Key features:
- Omnicell: A model for RNA-seq-based perturbation predictions.
- Benchmarking platform: A validation suite comparing performance across cell types and perturbations.
- Virtual Cell Atlas: Comprehensive in silico screening database featuring the largest collection of simulated cellular perturbations across 160+ diseases with single-cell data, allowing unprecedented scaling of biological insights.
By enabling researchers to simulate biological experiments computationally, we aim to accelerate the discovery of longevity therapeutics and optimize experimental strategies.
Opportunity
The Virtual Cell foundation model will:
- Enable rapid drug discovery by predicting effective interventions in aging-related diseases.
- Reduce experimental costs by prioritizing high-potential drug candidates before wet-lab testing.
- Improve reproducibility through a standardized framework for modeling cellular perturbations.
- Foster collaboration by providing an open-access resource for the longevity research community.
Budget
Funding request: $50,000 USD from VitaDAO, contributing fully to compute through contributing it to the decentralized open-source training run powered by Prime Intellect. The grant is in compute and might bring tokens in case of a retroactive reward.
Relevance to Longevity
Aging is driven by complex, multi-factorial processes at the cellular level. By enabling high-throughput simulations of gene regulation and drug interactions, the Virtual Cell model will:
- Identify novel longevity-promoting interventions.
- Enhance our understanding of cellular aging mechanisms.
- Provide a scalable tool for predicting the effects of therapeutics across diverse cell types.
IP & Open Science Roadmap
- Open-access model & datasets – The base version of the Virtual Cell model will be made publicly available for research purposes.
- Early-access partnerships – Select longevity-focused scientists will gain access to pre-release and enhanced versions.
- Potential commercialization – Development of premium AI-driven drug discovery tools for biotech and pharmaceutical partners.
Next Steps
- Secure funding from VitaDAO and additional partners.
- Develop and refine the Virtual Cell foundation model.
- Launch a pilot study with select longevity researchers.
- Publish findings and release the open-access dataset.
This initiative aligns with VitaDAO’s mission to drive innovation in longevity research by leveraging AI and computational biology. We invite the community to support and participate in the development of this transformative technology.
- Agree
- Revisions Requested (Detail in Comments)
- Disagree