An order-preserving batch-effect correction method based on a monotonic deep learning framework.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) reveals cellular heterogeneity and gene regulation.
• Batch-effect correction is critical for integrating scRNA-seq datasets.
• Existing procedural methods often overlook order preservation.

Purpose of the Study:

• To develop a novel batch-effect correction method for scRNA-seq data.
• To incorporate order preservation as a key feature in the correction process.
• To evaluate the method's performance against existing approaches.

Main Methods:

• Developed a monotonic deep learning network for batch-effect correction.
• Implemented an order-preserving feature within the deep learning framework.
• Compared the proposed method with established batch-correction techniques.

Main Results:

• The proposed method significantly improved clustering performance in scRNA-seq data.
• The method effectively retained original inter-gene correlation information.
• Differential expression information was better preserved compared to existing methods.

Conclusions:

• The novel order-preserving deep learning method enhances scRNA-seq data integration.
• This approach offers improved accuracy in analyzing cellular heterogeneity and gene regulation.
• The method addresses a key limitation in current procedural batch-correction techniques.