scAMACE

scAMACE performs joint modeling and clustering of single-cell chromatin accessibility, gene expression, and methylation to integrate multimodal data for characterizing cellular heterogeneity.

Key Features:

• Multimodal Integration: scAMACE integrates chromatin accessibility, gene expression, and methylation data from single cells into a joint analysis.
• Clustering Capabilities: scAMACE employs clustering methods to identify and characterize unknown cell types across modalities.
• Model-Based Joint Analysis: scAMACE uses a model-based framework to perform joint analysis of the three data types for holistic characterization of cellular heterogeneity.
• Expectation-Maximization Inference: scAMACE implements an Expectation-Maximization (EM) algorithm for statistical inference and parameter estimation.
• GPU Scalability: scAMACE provides a GPU implementation to increase computational efficiency and scalability for large single-cell datasets.
• Software Implementations: scAMACE is implemented in Python with CPU and GPU versions and has an R implementation.

Scientific Applications:

• Cell Type Identification: scAMACE enables identification and characterization of cell types by integrating multimodal signals.
• Development and Differentiation Studies: scAMACE supports analysis of cell differentiation and developmental processes at single-cell resolution.
• Disease Mechanisms and Downstream Analysis: scAMACE facilitates investigation of disease mechanisms and downstream analyses including pathway analysis and functional annotation.

Methodology:

Joint modeling of single-cell chromatin accessibility, gene expression, and methylation using a model-based framework with clustering and parameter estimation via an Expectation-Maximization (EM) algorithm; implementations in Python (CPU and GPU) and R.