Analysis updated 2026-05-18
Compare different spatial deconvolution methods on the same benchmark data.
Evaluate domain detection methods using a shared set of metrics.
Generate synthetic spatial transcriptomics data with SynthST for testing new methods.
| zafar-lab/spddb | abdurrafey237/rag-chatbot | humancompatibleai/pareto | |
|---|---|---|---|
| Stars | 3 | 3 | 3 |
| Language | Jupyter Notebook | Jupyter Notebook | Jupyter Notebook |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 4/5 | 3/5 | 2/5 |
| Audience | researcher | general | researcher |
Figures from each repo's GitHub metadata at analysis time.
Requires setting up separate conda environments per benchmarking method from provided yml files.
spDDB is a research toolkit for benchmarking methods used in spatial transcriptomics, a type of biology experiment that measures gene activity while keeping track of where in a tissue each measurement came from. The repository focuses on two computational tasks: spatial deconvolution, which estimates what mix of cell types is present at each spot in a tissue sample, and domain detection, which groups spots into regions that share similar biological characteristics. The project provides everything needed to run a fair comparison across methods. It includes conda environment files for each benchmarking method so researchers can install matching dependencies without conflicts, a synthetic data generator called SynthST for creating artificial spatial transcriptomics data and cell type proportions, and a collection of evaluation metrics covering bivariate spatial relationships, the shapes formed by different cell types, and rare cell type detection. The repository also links to a companion website hosting synthetic datasets that can be downloaded separately, and it includes a real dataset repository spanning tissues from brain, cancer, and other organs, collected across different species and spatial transcriptomics technologies. To get started, users clone the repository and create a conda environment from a provided yml file, first for SynthST and then for whichever benchmarking method they want to test, activating each environment before running it. The project comes from the Zafar Lab and is described in an accompanying research paper, credited to Ajita Shree, Aditya V, Tanush Kumar and Hamim Zafar. Contributions are welcome through GitHub issues for bug reports and questions, or through forking the repository and submitting a pull request for larger changes. The README does not state a specific license for the code.
A benchmarking toolkit that compares methods for estimating cell types and detecting tissue regions in spatial transcriptomics data.
Mainly Jupyter Notebook. The stack also includes Python, Jupyter Notebook, Conda.
The README does not state a license, so usage rights are unclear.
Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.
Mainly researcher.
This repo across BitVibe Labs
Verify against the repo before relying on details.