Towards
Plant Predictome

The Single-cell Foundation model Land

our Data

The Single-cell root Atlas

The annotated Arabidopsis root scRNA-seq atlas provides a foundation for annotating time-series data aimed at uncovering brassinosteroid signaling networks, and can be further expanded to include new avenues such as cell-cycle progression inference.

Shahan, R.*, C.-W. Hsu*, T.M. Nolan, B. Cole, I. Taylor, L. Greenstreet, S. Zhang, A. Afanassiev, A. Hendrika C. Vlot, G. Schiebinger, P.N. Benfey, and U. Ohler. (2022). A single-cell Arabidopsis root atlas reveals developmental trajectories in wild-type and cell identity mutants. Developmental Cell 57 (4), 543-560. e9 https://doi.org/10.1016/j.devcel.2022.01.008.
Hsu, C.-W., R. Shahan, M. Nolan, P.N. Benfey, and U. Ohler. (2022). Protocol for fast scRNA-seq raw data processing using scKB and non-arbitrary quality control with COPILOT. STAR Protocols 3, 101729. https://doi.org/10.1016/j.xpro.2022.101729.
Nolan, T.M.*, Vukasinovic*, C.-W. Hsu*, J. Zhang, I. Vanhoutte, R. Shahan, I. Taylor, L. Greenstreet, M. Heitz, A. Afanassiev, P. Wang, P. Szekely, A.Brosnan, Y. Yin, G. Schiebinger, U. Ohler, E. Russinova and P.N. Benfey. (2023). Brassinosteroid gene regulatory networks at cellular resolution in the Arabidopsis root. Science. https://doi.org/10.1126/science.adf4721.
Vukašinović N.*, C.-W. Hsu*, M. Marconi, S. Li, C. Zachary, R. Shahan, P. Szekley, Z. Aardening, I. Vanhoutte, Q. Ma, L. Pinto, P. Krupař, N. German, J. Zhang, C. Simon, J. Perez-Sancho, P. C. Quijada, Q. Zhou, L.R. Lee, J. Cai, E.M. Bayer, M. Fendrych, E. Truernit, Y. Zhou, S. Savaldi-Goldstein, K. Wabnik, T.M. Nolan, E. Russinova (2025). Polarity-guided uneven mitotic divisions control brassinosteroid activity in proliferating plant root cells. Cell 188 (8), 2063-2080. e24. https://doi.org/10.1016/j.cell.2025.02.011.

The rice single-cell root atlas

The annotated scRNA-seq atlas of rice (Oryza sativa) roots enables consistent annotation across conditions and reciprocally informs spatial transcriptomic data.

Cross modality & cross species integration

My research focus is to build multi-modal foundation models to understand and engineer plant resilience. Starting with large Arabidopsis single-cell RNA-seq corpora, we pretrain models and fine-tune them to predict how gene perturbations shape organ development and function. We then fuse spatial transcriptomics, metabolite profiles, and live imaging to link gene expression to cell size, shape, location, cell-cycle state, mechanical forces, and organ-level growth. By incorporating multi-omic data, we aim to uncover spatial-temporal “regulatory knobs” for precise control. In parallel, we tackle cross-species transfer by expanding high-quality crop genomes, leveraging DNA foundation models and refining cross-species analogous macrogene groupings with transcriptomic similarity, genome synteny, and protein function. Using Arabidopsis and rice as testbeds, we aim to deliver a robust genome and macrogene resource that carries predictions from model plants to crops, turning foundational insight into resilient agriculture.

Agentic AI powered Predictome

We are moving from static foundation models to agentic AI that can plan, reason, and act toward goals. Our agents will continuously gather and curate new data, run analyses, train and update models, and integrate results across modalities and species. Powered by strong foundation models, these systems use memory, planning, and action tools to operate end to end in silico, keeping our models current and delivering predictions that adapt to changing conditions. This agentic pipeline will scale plant science into a living, self-improving predictome.

Towards Plant Predictome