Xiang Lin | Biomedical Information & Computational Biology & AI

About

I develop AI models for denoising, processing, and analyzing single-cell multiomics and spatial transcriptomics data, as well as for advancing drug discovery.

Currently, I am a Postdoctoral Researcher jointly affiliated with Harvard Medical School and MIT. My long-term vision is to develop AI-driven platforms for precision medicine, combining advanced computational models with experimental technologies systems to accelerate therapeutic discovery.

I am passionate about building community resources for single-cells omics and therapeutic discovery, contributing to open-source projects, and collaborating across disciplines to push the frontier of computational biology and AI in medicine.

Research Interests

Single-cell and spatial transcriptomics analysis
Multimodal deep learning and generative models
AI methods for perturbation modeling and drug discovery
Benchmarking frameworks for reproducible bioinformatics
Integration of AI with experimental technologies (e.g., organ-on-a-chip)

Featured Projects

PDGrapher – Combinatorial prediction of therapeutic perturbations
CONCERT – Spatial-aware perturbation prediction
scMDC – Single-cell multi-omics deep clustering
DSSC – Constrained deep clustering for single-cell & spatial data
Multi-SC – Single-cell multi-omics pipeline
Benchmarking-Single-Cell-Perturbation – Resources for perturbation modeling

Selected Publications

PDGrapher: Combinatorial prediction of therapeutic perturbations using causally-inspired neural networks. Nature Biomedical Engineering, 2025 [Paper].

scMDC: Clustering of single-cell multi-omics data with multimodal deep learning. Nature Communications, 2022 [Paper].

DSSC: Model-based constrained deep clustering for spatial/single-cell data. Genome Research, 2022 [Paper].

MultiSC: Deep learning pipeline for multi-omics NEAT-seq analysis. Briefings in Bioinformatics, 2024 [Paper].

scAL: An active learning approach for clustering single-cell RNA-seq data. Lab. Invest., 2024 [Paper].

scDILT: scDILT: A Model-Based and Constrained Deep Learning Framework for Single-Cell Data Integration, Label Transferring, and Clustering. TCBB, 2025 [Paper].

Contact

Email: xiang_lin@hms.harvard.edu

GitHub: github.com/xianglin226