miQC: An adaptive probabilistic framework for quality control of single-cell RNA-sequencing data

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables high-resolution gene expression profiling.
• Quality control (QC) is crucial for removing low-quality cells before downstream analysis.
• Current QC methods independently use mitochondrial DNA (mtDNA) read proportion and gene detection counts, often with arbitrary thresholds, which can be overly stringent for certain tissues.

Purpose of the Study:

• To develop a data-driven quality control metric for scRNA-seq data.
• To jointly model mitochondrial DNA read proportion and gene detection for improved low-quality cell identification.
• To create a flexible QC method adaptable to diverse scRNA-seq datasets.

Main Methods:

• Proposed miQC, a novel data-driven QC metric.
• Utilized mixture models within a probabilistic framework to jointly analyze mtDNA read proportion and detected gene counts.
• Developed a software package for implementing the miQC metric.

Main Results:

• The miQC metric effectively identifies low-quality cells by jointly modeling key QC indicators.
• Demonstrated adaptability of miQC across different scRNA-seq datasets.
• miQC successfully removes low-quality cells while preserving high-quality cells for downstream analysis.

Conclusions:

• miQC offers a more accurate and less stringent approach to scRNA-seq quality control compared to current methods.
• This data-driven approach enhances the reliability of scRNA-seq analyses, particularly for challenging sample types.
• The miQC software package facilitates the application of this improved QC strategy.