Mdwgan-gp: data augmentation for gene expression data based on multiple discriminator WGAN-GP.

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Gene expression data is crucial for biological and medical research but is difficult and expensive to obtain experimentally.
• Generating high-quality gene expression data computationally is an urgent need.
• Existing methods like WGAN-GP can suffer from mode collapse or overfitting with small datasets.

Purpose of the Study:

• To address the limitations of current gene expression data augmentation techniques.
• To propose an improved generative adversarial network model for enhancing gene expression data.

Main Methods:

• Developed MDWGAN-GP, a generative adversarial network with multiple discriminators.
• Integrated a novel linear graph convolutional network for enriching training samples.
• Conducted extensive experiments using real biological data.

Main Results:

• MDWGAN-GP demonstrated superior performance in generating gene expression data.
• The method effectively augmented gene expression datasets, improving data quality.
• Experimental results validated the efficacy of the proposed approach.

Conclusions:

• MDWGAN-GP significantly enhances the quality of generated gene expression data.
• The proposed method outperforms state-of-the-art techniques in most cases.
• This approach offers a promising computational solution for gene expression data augmentation.