Developed a proprietary multimodal single-cell profiling technology and designed TM-Perturb-seq, a high-throughput sequencing platform.
First to collect aligned data on cell morphology, in situ sgRNA, chromatin, and transcriptome at single-cell resolution.
Used to fine-tune models for specific applications.

Real-World Data Network: Building dynamic clinical cohorts across multiple high-incidence cancers to create a world-leading, highly compliant real-world patient data network with deep longitudinal coverage.
Multimodal AI Foundation Model: Deeply integrating multidimensional spatiotemporal omics, including genomics, spatial transcriptomics, and proteomics, with clinical phenotype data to build a large-scale virtual tumor model.
Panoramic Microenvironment Simulation: Advancing from single tumor cell types to the complex tumor microenvironment (TME) to fully map cancer initiation, progression, and therapeutic escape.

Automated cleaning, annotation, and quality control of global single-cell data.
Cross-species, multimodal cell data (transcriptome, epigenome, and spatial transcriptome).
Standardized, high-quality datasets.

AI-Ready Data Factory

Aggregating high-quality single-cell, multimodal, and perturbation data from around the world to build a standardized, training-ready, and continuously evolving data foundation.