Yusuf Roohani

About:

I design new machine learning approaches for modeling biological systems. In particular, I'm interested in how artificial intelligence can inform better experiment design for discovering new medicines. My recent work has focused on computationally guiding the engineering of cells using genetic perturbations.

I'm currently a PhD student at Stanford University advised by Jure Leskovec and Stephen Quake. I'm affiliated with the Department of Biomedical Data Science, the Stanford AI Lab and the Stanford Machine Learning Group.

I previously worked as a machine learning engineer at GSK where I led a cross-disciplinary team to biologically profile their 2M+ compound collection using complex multi-modal datasets and high throughput screening. I enjoyed the four years that I spent in industry designing robust machine learning systems that fit a healthcare context - from discovery all the way to diagnostics.

Curriculum Vitae:

[CV]

Featured Publications:

For complete list: Google Scholar

Journal Publications

Nature Methods [2023] Towards Universal Cell Embeddings: Integrating Single-cell RNA-seq Datasets across Species with SATURN
Rosen Y.*, Brbic M.*, Roohani, Y.*, Swanson K., Li Z., Leskovec, J..

Nature Biotechnology [2023] [Code] GEARS: Predicting transcriptional outcomes of novel multi-gene perturbations
Roohani, Y., Huang, K., Leskovec J.
    Oral presentations and Awards:
     [Best Poster] Intelligent Systems For Molecular Biology (ISMB 2022)
     [Innovation Award] Society for Lab Automation and Screening (SLAS 2023)
     [Best Poster] [Video] Single-Cell Genomics Meets Data Science 2022
     [Video] Machine Learning for Computational Biology 2022 (MLCB2022)
     CRISPR Perturbations and Beyond 2022 (Wellcome Sanger Institute)
     [GitHub 150+ Stars]

(* = equal contribution)

Preprints

arXiv [2023] [Code] Universal Cell Embeddings: A Foundation Model for Cell Biology
Rosen Y.*, Roohani Y.*, Agrawal A., Samotorcan L., Quake S., Leskovec J..
[Media coverage: New York Times]
[Stanford BioX Interdiscplinary Initiatives Poster Award]

arXiv [2022] CausalBench: A Large-Scale Benchmark for Network Inference from Single-Cell Perturbation Data
Chevalley, M., Roohani, Y., Mehrjou, A., Leskovec, J., Schwab, P..

Conference Papers

ICLR [2024]: LLMs for Agents Workshop, MLGenX Workshop
BioDiscoveryAgent: An AI agent for designing genetic perturbation experiments
Roohani Y.*, Vora J.*, Huang Q.*, Steinhart Z., Marson A., Liang P., Leskovec J..
[Spotlight Presentation]

NeurIPS [2023] Zero-shot causal learning
Nilforoshan H.*, Moor M.*, Roohani Y., Chen Y., Surina A., Yasunaga M., Oblak S., Leskovec J..
[Spotlight Presentation]