or4k2l/scVAE-Annotator
Advanced Single-Cell RNA-seq Annotation Pipeline with VAE and Automated Hyperparameter Optimization
What's novel
scVAE-Annotator is an optimized pipeline for automated cell type annotation in single-cell RNA-seq data. It combines in AI cooperation
Code Analysis
12 files read · 3 roundsA single-cell RNA-seq annotation pipeline that trains a VAE for dimensionality reduction, performs Leiden clustering, and uses calibrated ML classifiers (XGBoost/LogReg/SVC) with Optuna hyperparameter optimization to predict cell types with confidence scores.
Strengths
Well-structured pipeline with scientifically-grounded VAE implementation (Poisson/MSE likelihoods, KL warm-up, numerical stability), comprehensive evaluation framework with calibration plots, and clean separation of concerns across modules.
Weaknesses
Test coverage is only 31%, some legacy code paths remain (minimal Annotator class), and the pipeline depends on external data downloads which could be fragile in production environments.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
66
Contributors
2
Files
74
Active weeks
5
Repository
Language
Python
Stars
0
Forks
0
License
MIT