Skip to main content

Development Branch

StatescopePro (Dev) — Overview

StatescopePro is the next-generation (development) implementation of Statescope: a memory- and time-efficient PyTorch-based rewrite designed for faster, more scalable deconvolution on large cohorts—while preserving the same core Bayesian log-normal model.

Where to find it​

You can find StatescopePro on GitHub in the development branch:

  • Repo: tgac-vumc/Statescope
  • Branch: dev
  • Folder / module: StatescopePro

The StatescopePro directory contains a README with the most up-to-date API usage, installation notes, and troubleshooting steps.

  • PyTorch rewrite: faster runtimes + easier GPU use.

  • Lower memory: more efficient tensors/caching → better for large cohorts.

  • Same model: preserves the Bayesian log-normal framework; results stay near-identical (within numerical tolerance).

  • Cleaner codebase: easier profiling, debugging, and extending.