scMultiGAN: cell-specific imputation for single-cell transcriptomes with multiple deep generative adversarial...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables high-resolution cellular analysis but suffers from missing data.
• Existing imputation methods often fail due to oversimplified assumptions about gene expression distributions.
• Accurate imputation is crucial for downstream scRNA-seq data analyses.

Purpose of the Study:

• To develop a novel deep learning model for accurate imputation of missing values in scRNA-seq data.
• To address the limitations of current imputation techniques by considering cell-specific gene expression patterns.
• To provide a scalable and robust imputation solution for diverse scRNA-seq datasets.

Main Methods:

• Development of scMultiGAN, a deep learning model employing multiple collaborative generative adversarial networks (GANs).
• Implementation of a two-stage training process for cell-specific imputation.
• Evaluation of scMultiGAN against state-of-the-art methods using imputation accuracy, cell clustering, differential gene expression, and trajectory analysis.

Main Results:

• scMultiGAN demonstrated superior imputation accuracy compared to existing methods.
• The model significantly improved performance in downstream analyses including cell clustering, differential gene expression, and trajectory inference.
• scMultiGAN proved scalable for large scRNA-seq datasets and robust across different sequencing platforms.

Conclusions:

• scMultiGAN offers a powerful and effective solution for scRNA-seq data imputation.
• The cell-specific imputation approach enhances the reliability of scRNA-seq data analysis.
• The developed model represents a significant advancement in single-cell data processing.