Hist2ST

JHU EN.580.697 26 fall, Project F. A model that is trained to predict spatial gene expression (Spatial Transcriptomics) from H&E stains (Histogram). Data comes from HEST-1k.

How to run

Using venv

uv venv --python 3.11
pip install -r requirements.txt

Using conda (conda needs to be installed first)

In a conda terminal

conda create -n my_env_name python=3.11
conda activate my_env
pip install -r requirements.txt

Document Structure

• .env the huggingface access key. NEVER COMMITTED
• .env.example a sample of what an .env would look like.
• README.md (this file) what the project is, how to install, how to run
• requirements.txt dependencies
• .gitignore includes the data folder data/, output folder outputs/, checkpoints checkpoints/, large files
• configs/ experiment settings (e.g., YAML)
• src/ reusable code: data loading, models, metrics
• scripts/ short helpers like train.py, evaluate.py, make_figures.py
• notebooks/ EDA and demos (call functions from src/)
• splits/ saved train/validation/test ID lists
• jobs/ cluster batch scripts
• docs/ notes, meeting notes, design choices
• Presentation/ presentation slides

In .gitignore: Large data, checkpoints (checkpoints), and generated outputs (outputs) are not tracked in Git