Metadata-Version: 2.4
Name: scDecant
Version: 0.1.1
Summary: Label-free, batch-aware feature selection for single-cell multi-omics
Author: Ziyi Zou, Heyang Hua, Siyu Li, Yuhang Jia
Author-email: Shengquan Chen <chenshengquan@nankai.edu.cn>
License: MIT
Project-URL: Repository, https://github.com/BioX-NKU/scDecant
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: numpy>=1.23
Requires-Dist: pandas>=1.5
Requires-Dist: scipy>=1.9
Requires-Dist: scikit-learn>=1.1
Requires-Dist: scanpy>=1.9
Requires-Dist: anndata>=0.8
Requires-Dist: scib>=1.1
Provides-Extra: tutorials
Requires-Dist: episcanpy>=0.4; extra == "tutorials"
Requires-Dist: matplotlib>=3.5; extra == "tutorials"
Requires-Dist: jupyter; extra == "tutorials"
Dynamic: license-file

As single-cell transcriptomic and epigenomic atlases expand across samples, protocols, and laboratories, batch effects can confound biological structure. Although downstream methods can mitigate or quantify batch effects, they are applied only after feature selection has fixed the input space, whereas most feature selectors retain or discard features without explicitly balancing biological signal against batch association. Here we introduce DECANT, a label-free feature selector that weighs biological discrimination against batch association for each feature. Across single-cell RNA sequencing (scRNA-seq), single-cell DNA methylation (scDNAm), and single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq) data, this batch-aware ranking consistently improved integration performance while preserving interpretable regulatory and epigenetic signals. Beyond integration benchmarks, features selected by DECANT supported regulatory pathway analysis at cell type resolution, enabled the identification of more candidate differentially methylated regions, and exhibited stronger specificity for brain tissues in cortical methylation data.
