scGNN 2.0

scGNN 2.0 applies graph neural network methodologies to analyze single-cell RNA sequencing (scRNA-Seq) data for gene expression imputation and cell clustering.

Key Features:

• Superior Performance: A streamlined architecture accelerates processing times and scales to large scRNA-Seq datasets while maintaining analytical accuracy.
• Improved Imputation and Clustering: Reduces Median L1 Error by 67.94% on average across eight datasets and increases adjusted rand index by 85.02% on average for clustering performance.
• Visualizations and Interpretability: Produces visualization outputs to assess imputation quality, compare against benchmarks, and interpret cell–cell interaction patterns.
• Expanded Compatibility: Supports a broader range of input and output formats and integration with single-cell analysis toolkits in Python and R environments.

Scientific Applications:

• Cellular heterogeneity analysis: Uncovers cellular heterogeneity in single-cell transcriptomic datasets.
• Gene expression imputation: Improves gene expression imputation in scRNA-Seq data to recover missing or noisy measurements.
• Cell clustering and biological interpretation: Enhances cell clustering to support cell-type identification and elucidation of complex biological processes.

Methodology:

Applies graph neural network methodologies for gene expression imputation and cell clustering and evaluates performance using Median L1 Error and adjusted rand index across datasets.