FEAST
Simulation and interpolation of spatial transcriptomics from parameter cloud
FEAST: Simulation and interpolation of spatial transcriptomics from parameter cloud
Status: Independent first-author project; manuscript in final preparation
Role: First author, Independent Research Project
Institution: ZJU-UoE Institute, Zhejiang University; collaboration with Prof. Maizie Zhou Lab, Vanderbilt University
Duration: August 2024 – Present (Remote and on-site collaboration)
Project Overview
FEAST (FEAture-space based modeling for Spatial Transcriptomics) is an independent first-author project on parameter-cloud modeling for spatial transcriptomics. The framework represents gene-level mean, variance, and sparsity in a latent parameter space, then uses that structure to simulate realistic slices and interpolate tissue states across space.
Beyond two dimensions, FEAST extends to 3D interpolation through optimal-transport-guided transitions between parameter clouds, supporting volumetric reconstruction and systematic benchmarking for spatial omics methods.
Technical Innovation
Statistical Modeling:
- Parameter-cloud representation: models gene-level mean, variance, and sparsity in a unified latent space
- C-vine copula modeling: captures nonlinear and asymmetric dependence among gene parameters
- Flexible count models: supports Poisson, Negative Binomial, ZIP, and ZINB distributions
Simulation Framework:
- Single-slice simulation: generates realistic 2D spatial transcriptomics slices with controllable perturbations
- Expression alteration: tunes mean, variance, and sparsity for robustness testing
- Spatial transformation: enables geometry-aware benchmarking for alignment methods
3D Interpolation:
- Optimal transport: Wasserstein-style interpolation between parameter clouds
- Alignment-guided coordinates: transport-informed coordinate interpolation
- Volumetric reconstruction: generates intermediate slices for continuous 3D tissue architectures
Key Results
- High-fidelity simulation with near-perfect correlation for gene means and variances across multiple ST platforms (10x Visium, MERFISH, OpenST, Slide-seqV2, Xenium, Stereo-seq)
- Clustering benchmarking revealed algorithmic sensitivities under controlled expression and sparsity perturbations
- Alignment evaluation demonstrated Spateo’s robustness over SPACEL under geometric transformations
- 3D interpolation achieved >0.9 correlation with ground-truth experimental slices in leave-one-out validation
Research Scope
This project has involved the full cycle of independent method development:
- literature review and problem formulation
- model construction and implementation
- benchmarking and application studies
- manuscript preparation and figure development
Current Positioning
FEAST is the clearest expression of my independent research direction so far: building mathematically grounded computational frameworks for simulation, benchmarking, and representation of spatial biological data.