Scalable preprocessing for sparse scRNA-seq data exploiting prior knowledge.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) generates vast datasets requiring sophisticated computational analysis.
• Existing tools struggle with data sparsity, large scale, and integrating prior biological knowledge.

Purpose of the Study:

• To introduce UNCURL, a novel preprocessing framework for scRNA-seq data.
• To enhance the analysis of scRNA-seq data by addressing current computational challenges.

Main Methods:

• Utilizes non-negative matrix factorization (NMF) for scRNA-seq data preprocessing.
• Designed to handle varying sampling distributions and large cell numbers.
• Incorporates prior biological knowledge, such as bulk RNA-seq or marker information.

Main Results:

• UNCURL preprocessing improves clustering, visualization, and lineage estimation in scRNA-seq analysis.
• Demonstrates consistent performance improvement with and without prior knowledge integration.
• Exhibits high scalability and parallelizability, outperforming other methods on large datasets (1.3 million cells).

Conclusions:

• UNCURL offers a robust and scalable solution for scRNA-seq data preprocessing.
• The framework effectively integrates prior biological information to enhance data analysis.
• UNCURL significantly advances the computational analysis of large-scale single-cell transcriptomic data.