seqQscorer: automated quality control of next-generation sequencing data using machine learning

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Next-generation sequencing (NGS) generates vast amounts of data, necessitating robust quality control measures.
• Assessing the quality of NGS data files is crucial for reliable downstream analysis but remains a complex challenge.
• Existing quality control methods may not fully capture the nuances of diverse NGS data types.

Purpose of the Study:

• To develop and validate a novel, automated quality control procedure for next-generation sequencing data.
• To statistically characterize common quality features in NGS data files.
• To create generalizable predictive models for assessing NGS data quality across different species and experimental contexts.

Main Methods:

• Statistical characterization of common quality control features in NGS data.
• Development of classification algorithms, including tree-based and deep learning approaches.
• Validation of predictive models using internal and external functional genomics datasets.

Main Results:

• The developed models demonstrate effectiveness in statistically characterizing NGS quality features.
• Predictive models show generalizability to data from unseen species, indicating robustness.
• The quality control procedure successfully automates the assessment of NGS data quality.

Conclusions:

• The proposed statistical guidelines and predictive models offer a valuable resource for NGS data users.
• Automated quality control using machine learning can significantly improve the understanding and management of NGS data quality.
• The developed software and guidelines facilitate better interpretation of quality issues and enhance data reliability.