TemSOMap
Mapping lineage-resolved scRNA-seq data with spatial transcriptomics
TemSOMap: Mapping lineage-resolved scRNA-seq data with spatial transcriptomics
Status: Collaborative manuscript in final preparation
Role: Research Assistant, Co-author
Institution: ZJU-UoE Institute, Zhejiang University; collaboration with Prof. Xiuwei Zhang Lab, Georgia Tech
Duration: April 2024 – September 2025 (Remote collaboration)
Project Overview
TemSOMap addresses a core problem in spatial transcriptomics: how to map lineage-resolved single-cell RNA-seq profiles back onto spatial measurements. The project develops a lineage-aware computational framework for inferring cell-to-spot assignments while respecting both expression similarity and developmental structure.
Technical Approach
The methodology combines machine learning and structured optimization:
Core Algorithm:
- Loss-function design: balances expression agreement with lineage-aware constraints
- Cell-to-spot mapping: infers probabilistic assignment matrices between single cells and spatial spots
- Lineage integration: incorporates developmental information into spatial reconstruction
Machine Learning Framework:
- expression-based similarity metrics
- lineage-aware regularization
- scalable optimization for large mapping problems
My Contributions
My contributions focused on:
- downstream analysis and biological validation of mapping results
- partial benchmark development for method comparison
- figure preparation, method comparison, and manuscript support
Current Positioning
TemSOMap helped shape my interest in computational frameworks that connect multiple biological views of the same system. Related lineage-aware work from this collaboration also contributed to the subsequent LineageMap project, which has been accepted at RECOMB 2026 and is expected to appear in Genome Research.