Screen technical noise in single cell RNA sequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (SC RNA-seq) is a powerful technology for analyzing cellular heterogeneity.
• Technical noise and batch effects can significantly impact the accuracy of SC RNA-seq data analysis.
• Robust data cleaning methods are crucial for reliable interpretation of SC RNA-seq experiments.

Purpose of the Study:

• To develop and validate a comprehensive data cleaning pipeline for SC RNA-seq data.
• To address technical noise and library variability in SC RNA-seq datasets.
• To provide an accessible R package implementation of the proposed data cleaning pipeline.

Main Methods:

• A two-stage data cleaning pipeline involving gene-wise and library-wise screening.
• Gene-wise screening utilizes negative binomial regression to assess gene count trends against library size.
• Library-wise screening compares correlations of housekeeping (HK) genes versus non-housekeeping (NHK) genes across libraries.

Main Results:

• The proposed pipeline effectively screens genes and cell libraries in SC RNA-seq data.
• The method successfully identified and removed low-quality libraries based on HK and NHK gene correlations.
• The pipeline was successfully applied to two large-scale SC RNA-seq datasets, demonstrating its scalability and effectiveness.

Conclusions:

• The developed data cleaning pipeline offers a significant improvement in SC RNA-seq data quality.
• The R package provides a user-friendly tool for researchers to implement robust data preprocessing.
• This approach enhances the reliability of downstream analyses and biological discoveries from SC RNA-seq data.