SPATIA

Getting Started

  • Installation
    • Conda Environment
    • Package Installation
    • Generation Dependencies
    • Notes

Data & Training

  • MIST Dataset Construction
    • Stage A: Crop Cell Images into LMDB
    • Stage B: Build Annotated h5ad and Register in lamindb
    • Stage C: Merge Per-Dataset LMDBs
    • Data Loading at Training Time
  • Representation Training (Stage 1)
    • Prerequisites
    • SPATIA-scprint (Primary)
      • Configure paths
      • Run training
      • Extract embeddings (SPATIA-scprint)
    • SPATIA-scgpt (Alternative)
      • Run training
      • Extract embeddings (SPATIA-scgpt)
    • Architecture Summary

Downstream Tasks

  • Prediction Tasks
    • Table 2: Cross-Platform Clustering (Xenium + CosMx)
      • Step 1 — Extract SPATIA embeddings
      • Step 2 — Run multi-seed clustering
    • Table 3: Biomarker Prediction (HEST Benchmark)
    • Table 4: scRNA-seq Clustering & Annotation (GSE155468)
      • Option A — SPATIA only
      • Option B — All baselines
      • Cell Annotation
    • Labels and Reproducibility
  • Generation Pipeline (Stages 2 & 3)
    • Stage 2: OT-Based Perturbation Pairing
      • Cell-level pairing
      • Niche-level pairing (grid-based)
    • Stage 3: Flow-Matching Image Generation
      • Pretrained Checkpoints
      • Configure data paths
      • Training
      • Output structure
      • Evaluation (FID)
      • Morphology proxy encoder
    • Niche-Level Generation

Reference

  • Citation & Contact
    • Citation
    • Contact
SPATIA
  • Search

© Copyright 2025, Zitnik Lab, Harvard Medical School.

Built with Sphinx using a theme provided by Read the Docs.