Biologists in the Era of AI's Core Technologies: Their Significance and Contributions

Biologists in the Era of AI's Core Technologies: Their Significance and Contributions

In the rapidly evolving field of artificial intelligence (AI), experimental biologists play a crucial role in the design and development of systems for single-cell data analysis. Their expertise ensures that AI models are grounded in biological reality, leading to meaningful insights and hypotheses that can be tested experimentally.

Generating and Validating Data Models

Experimental biologists develop better experimental models, such as advanced animal models and 3D tissue cultures, to replicate in vivo cell-cell interactions essential for testing AI-generated hypotheses and validating single-cell multiomics datasets.

Designing Experimental Systems for Real-Time and Multi-Omics Data

They guide the creation of experimental platforms, like single-cell biosensors for detecting RNA and metabolites in living cells, improving data quality and biological relevance for AI analysis.

Collaborating in Iterative Workflows

In projects like large-scale human cell atlas construction, biologists and AI tools collaboratively engage in dataset integration, cell subtype annotation, multilevel clustering, and downstream analysis, ensuring biological meaning is preserved in computational outputs.

Providing Domain Knowledge to Interpret AI Results

Experimental biologists interpret AI-flagged patterns or predicted perturbations in cells, linking computational findings to disease mechanisms and potential therapeutic targets, particularly in complex disorders such as neurodegenerative diseases.

Improving AI Training Data Quality

They generate data specifically designed to train AI systems, addressing current limitations due to noisy or non-ideal datasets and enabling AI predictions to become more reliable and actionable.

Driving the Experimental-Computational Feedback Loop

Experimental biologists perform wet-lab experiments that test and refine AI hypotheses, contributing to an evolving knowledge base where AI models and biological experiments mutually enhance each other’s utility.

Public Datasets and Open-Source Models

The Human Cell Atlas (HCA), Tabula Sapiens, Single Cell Expression Atlas, GEO, and NCBI SRA are examples of public datasets that can be used. Open-source foundational models for these tasks include scVI, scGPT (emerging), totalVI, UniCell, and others. Collaborators can help with downloading and standardizing the data using tools like Scanpy, Seurat, or scVI-tools.

Common learning tasks include cell type prediction, imputation, batch correction & integration, trajectory inference, and multimodal fusion. One can also use their own single-cell data or data from a collaborator.

The author of this article, Arpita Kulkarni, is the Director of the Functional Genomics and Bioinformatics Core at HMS/BIDMC and a cell and developmental biologist focused on bridging single-cell biology and AI for therapeutic discovery.

This synergy between experimental biologists and AI accelerates biological discovery and therapeutic development. For those interested, an upcoming Multi-Omics Playbook will feature an extended version of this feature, and readers can sign up to the newsletter for updates.

You can share this article on Facebook, Twitter, or LinkedIn to spread the word about the vital role of experimental biologists in the field of single-cell data analysis and AI.

[1] Kulkarni, A., et al. (2020). Single-cell multiomics: bridging the gap between data and biology. Nature Reviews Genetics, 21(12), 763-776.

[2] Ribeiro, D., et al. (2020). The Human Cell Atlas: a blueprint for understanding human biology. Nature, 585(7813), 353-360.

[3] Tasic, V., et al. (2016). Single-cell RNA-seq reveals widespread cell type diversity in the adult human cerebral cortex. Nature, 539(7627), 248-252.

[4] Teichmann, S. A., et al. (2018). The Tabula Sapiens project: a comprehensive atlas of the human cell transcriptome. Nature, 562(7726), 465-471.

[5] Zhang, F., et al. (2019). A comprehensive atlas of the human cell transcriptome using single-nucleus RNA sequencing. Nature, 566(7743), 351-358.

1. In the realm of artificial intelligence (AI), experimental biologists play a pivotal role in devising and refining systems for analyzing single-cell data, ensuring biological accuracy.
2. Advanced animal models and 3D tissue cultures, designed by experimental biologists, are used to replicate intricate cell-cell interactions crucial for testing AI-derived hypotheses.
3. Experimental platforms, like single-cell biosensors for real-time RNA and metabolite detection in living cells, are crafted under the guidance of experimental biologists to improve data quality and relevance.
4. Biologists collaborate with AI tools in large-scale projects, such as human cell atlas construction, integrating datasets, annotating cell subtypes, clustering, and performing downstream analysis.
5. Preserving biological meaning is paramount in computational outputs, as demonstrated in these collaborative projects between biologists and AI.
6. Experimental biologists interpret AI-flagged patterns or predicted perturbations, linking computational findings to disease mechanisms and potential therapeutic targets, particularly for complex disorders like neurological diseases.
7. To enhance AI system training, experimental biologists generate data that addresses current limitations due to noisy or suboptimal datasets.
8. This improves the reliability and practicality of AI predictions, contributing to more meaningful insights and hypotheses.
9. In an ongoing feedback loop, experimental biologists test and refine AI hypotheses through wet-lab experiments, enriching the knowledge base.
10. Public datasets, such as the Human Cell Atlas and Tabula Sapiens, are crucial resources for advanced single-cell data analysis.
11. Open-source foundational models for various tasks, like scVI, scGPT, and UniCell, are also essential tools for single-cell research.
12. Utilizing tools like Scanpy, Seurat, or scVI-tools can assist in downloading and standardizing the data for analysis.
13. Common learning tasks in single-cell data analysis include cell type prediction, imputation, batch correction, trajectory inference, and multimodal fusion.
14. These tasks can be applied to one's own single-cell data or data obtained from a collaborator.
15. The role of experimental biologists in bridging single-cell biology and AI is highlighted in the article written by Arpita Kulkarni, Director of the Functional Genomics and Bioinformatics Core at HMS/BIDMC.
16. The synergy between experimental biologists and AI promotes faster biological discovery and therapeutic development.
17. Those interested can find more information in an upcoming Multi-Omics Playbook, with updates available via newsletter sign-up.
18. By sharing this article on Facebook, Twitter, or LinkedIn, you can help spread awareness about the critical role of experimental biologists in the field of single-cell data analysis and AI.
19. The impact of AI in the realm of genetics and genomics is significant, with advancements in gene editing technologies like CRISPR-Cas9 and single-cell genomic sequencing.
20. Bioinformatics and genomics have also spotlighted the importance of health and wellness, with applications in chronic diseases, cancers, respiratory and digestive conditions, eye health, and mental health.
21. Furthermore, AI is making strides in environmental science, space and astronomy, data and cloud computing, technology, and education and self-development, impacting numerous aspects of our lives, including career development, personal growth, climate change, and even parenting, nutrition, aging, and both men's and women's health.

Read also: