scAI

scAI performs unsupervised integration of single-cell gene expression with chromatin accessibility or DNA methylation to dissect cellular heterogeneity and transcriptional regulatory mechanisms.

Key Features:

• Unsupervised Integration: Integrates parallel single-cell transcriptomic and epigenomic profiles (chromatin accessibility or DNA methylation) using an unsupervised learning framework.
• Iterative Learning Process: Refines cellular similarity and signal aggregation through iterative learning to improve alignment between epigenomic and transcriptomic layers.
• Aggregation of Sparse Epigenomic Signals: Aggregates sparse epigenomic signals across similar cells to enhance signal-to-noise before fusion with gene expression data.
• Dissection of Transcriptional Regulation: Enables analysis of how chromatin accessibility and DNA methylation patterns influence single-cell gene expression to study regulatory mechanisms.

Scientific Applications:

• Single-Cell Multi-omic Analysis: Characterizes cellular states and heterogeneity by jointly analyzing single-cell transcriptomic and epigenomic measurements.
• Transcriptional Regulation Studies: Investigates relationships between chromatin accessibility, DNA methylation, and gene expression to elucidate regulatory programs and cell-fate decisions.

Methodology:

Use unsupervised learning to aggregate sparse epigenomic signals from similar cells with iterative refinement; fuse aggregated chromatin accessibility or DNA methylation data with single-cell gene expression; validate performance using simulations and real datasets.