plinkQC: An Integrated Tool for Ancestry Inference, Sample Selection, and Quality Control in Population Genetics.

Area of Science:

• Population genetics
• Bioinformatics
• Computational biology

Background:

• High-quality datasets are essential for accurate population genetic analyses.
• Existing software packages often lack integrated solutions for crucial preprocessing steps like ancestry identification and sample relatedness assessment.

Purpose of the Study:

• To develop a comprehensive R package, plinkQC, that integrates essential quality control and preprocessing steps for population genetic analyses.
• To provide a user-friendly tool that automates ancestry determination, sample relatedness pruning, and quality control checks.

Main Methods:

• Developed plinkQC, an R/CRAN package combining multiple functionalities.
• Implemented a pre-trained random forest classifier for accurate ancestry determination (98% accuracy with 5% marker overlap).
• Created a graph-based pruning method to select maximal sets of unrelated samples, considering relationship estimates and sample quality.

Main Results:

• plinkQC successfully integrates ancestry determination, sample relatedness pruning, and quality control into a single package.
• The random forest classifier achieves high accuracy in ancestry prediction.
• The graph-based pruning method identified additional samples compared to existing methods, as demonstrated on the 1000 Genomes project.
• The package provides detailed quality control reports and outputs cleaned datasets.

Conclusions:

• plinkQC offers a unified and efficient solution for critical population genetic data preprocessing.
• The package enhances the quality and size of sample datasets, facilitating more robust genetic studies.
• plinkQC is readily available on CRAN with comprehensive documentation and code on GitHub.