scCAN: Clustering With Adaptive Neighbor-Based Imputation Method for Single-Cell RNA-Seq Data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for analyzing cellular heterogeneity.
• Technical limitations in scRNA-seq lead to dropout events, causing zero gene expression values.
• Accurate imputation of these zero values is essential for reliable downstream analysis.

Purpose of the Study:

• To introduce scCAN, a new imputation method for scRNA-seq data.
• To address the challenge of dropout events in gene expression matrices.
• To enhance the accuracy of downstream analyses by improving dropout imputation.

Main Methods:

• scCAN utilizes adaptive neighborhood clustering to estimate zero gene expression values.
• The method iteratively refines cell-cell similarity, clustering structures, and Laplacian matrix constraints.
• A novel approach is employed to update similarity information for improved imputation.

Main Results:

• scCAN demonstrated superior performance in imputing dropout zero values across simulated and real scRNA-seq datasets.
• Downstream analyses, including cell clustering and trajectory reconstruction, showed significant improvement with scCAN.
• Comparative evaluations confirmed scCAN outperforms existing imputation methods.

Conclusions:

• scCAN offers a robust and effective solution for handling dropout events in scRNA-seq data.
• The method enhances the reliability and accuracy of scRNA-seq data analysis.
• scCAN is a valuable tool for researchers studying cellular heterogeneity.