Cellcano

Cellcano applies supervised learning to identify cell types from single-cell Assay for Transposase-Accessible Chromatin sequencing (scATAC-seq) data, enabling accurate annotation of cellular identities based on chromatin accessibility and epigenetic heterogeneity.

Key Features:

• Two-Round Supervised Learning Algorithm: Employs a two-round supervised learning procedure that mitigates distributional shift between reference (training) and target (prediction) scATAC-seq datasets to improve cell type prediction accuracy.
• Systematic Benchmarking: Evaluated across 50 well-designed cell typing tasks from diverse scATAC-seq datasets to assess accuracy, robustness, and computational efficiency.

Scientific Applications:

• Cell Type Identification: Accurate annotation of cell types from scATAC-seq profiles based on chromatin accessibility signals.
• Epigenetic Heterogeneity and Regulatory Mechanisms: Analysis of epigenetic heterogeneity and chromatin accessibility to support studies of cellular regulatory mechanisms.

Methodology:

Uses a two-round supervised learning algorithm to address distributional shift between reference and target scATAC-seq datasets and was systematically benchmarked across 50 cell typing tasks.