# How to Build the Virtual Cell with Artificial Intelligence: Priorities and Opportunities

**Authors:** Charlotte Bunne, Yusuf Roohani, Yanay Rosen, Ankit Gupta, Xikun Zhang, Marcel Roed, Theo Alexandrov, Mohammed AlQuraishi, Patricia Brennan, Daniel B. Burkhardt, Andrea Califano, Jonah Cool, Abby F. Dernburg, Kirsty Ewing, Emily B. Fox, Matthias Haury, Amy E. Herr, Eric Horvitz, Patrick D. Hsu, Viren Jain  
**arXiv:** [2409.11654](https://arxiv.org/abs/2409.11654)  
**Date:** 2024-09-18  
**Summary page:** https://www.summarizepaper.com/en/paper/2409.11654v1/

## Key points

- The cell is the fundamental unit of life, and understanding its complexities is crucial for unraveling the mysteries of biology and identifying disease causes.
- Recent advancements in artificial intelligence (AI) have enabled the modeling of cells by combining vast amounts of experimental data.
- AI-powered Virtual Cells are being developed to create robust representations directly learned from diverse biological data, integrating molecular interactions and signaling networks within a cell.
- Various measurements such as transcriptome analysis, chromatin accessibility, proteome profiling, and imaging technologies are used to create a comprehensive representation of the cell.
- AI algorithms like vision transformers and convolutional neural networks are applied to capture different biological features across multiple imaging channels.
- Autoencoders and transformers play a key role in learning representations for sequence-based data while integrating different modalities collected from sequencing and imaging technologies.
- Modeling cellular organelles and membraneless compartments as distinct units within the cell is essential for accurate predictions and mechanistic interpretability as AI Virtual Cells become more complex.
- Incorporating imaging modalities will continue to evolve the cellular representation to effectively model spatial organization and dynamics with open science collaborations across academia, philanthropy, biopharma industries, and AI sectors driving this vision forward.
- AI Virtual Cells hold immense potential in identifying new drug targets, predicting cellular responses to perturbations, scaling hypothesis exploration, and achieving a predictive understanding of cell mechanisms for broader utility in biomedical research.

## Summary

The cell is the fundamental unit of life and understanding its complexities is crucial for unraveling the mysteries of biology and identifying disease causes. Recent advancements in artificial intelligence (AI) have opened up new possibilities for modeling cells by combining vast amounts of experimental data. This has led to the concept of AI-powered Virtual Cells, where robust representations can be directly learned from diverse biological data. By integrating molecular interactions and signaling networks within a cell, individual cell states can be accurately modeled at the cellular scale. This involves incorporating various measurements such as transcriptome analysis, chromatin accessibility, proteome profiling, and imaging technologies to create a comprehensive representation of the cell. To capture different biological features across multiple imaging channels, AI algorithms like vision transformers and convolutional neural networks are applied to biological images. Autoencoders and transformers play a key role in learning representations for sequence-based data while integrating different modalities collected from sequencing and imaging technologies. As AI Virtual Cells become more complex, it becomes essential to also model cellular organelles and membraneless compartments as distinct units within the cell to ensure accurate predictions and mechanistic interpretability. The cellular representation will continue to evolve by incorporating imaging modalities to effectively model spatial organization and dynamics with open science collaborations across academia, philanthropy, biopharma industries, and AI sectors driving this vision forward. Ultimately, AI Virtual Cells hold immense potential in identifying new drug targets, predicting cellular responses to perturbations, scaling hypothesis exploration, and achieving a predictive understanding of cell mechanisms for broader utility in biomedical research.