SEQSpark: A Complete Analysis Tool for Large-Scale Rare Variant Association Studies Using Whole-Genome and Exome...

Area of Science:

• Genetics and Genomics
• Bioinformatics
• Computational Biology

Background:

• Large-scale sequencing studies are crucial for identifying rare variants associated with complex diseases.
• Current analytical tools struggle to efficiently process the massive datasets generated by these studies.
• There is a need for faster and more efficient computational methods to analyze whole-genome and exome sequence data.

Purpose of the Study:

• To develop and validate SEQSpark, a novel parallel processing tool for large-scale sequence data analysis.
• To demonstrate the speed and efficiency of SEQSpark in performing data quality control, annotation, and association analyses.
• To enable rapid elucidation of genetic variation involved in the etiology of complex traits.

Main Methods:

• SEQSpark implements parallel processing using Apache Spark for enhanced speed and efficiency.
• The tool was tested on whole-genome sequence data from the UK10K cohort for association analysis with waist-to-hip ratio.
• Performance was benchmarked against existing tools like Variant Association Tools and PLINK/SEQ using simulated and real data.

Main Results:

• SEQSpark successfully analyzed >9 million variants from UK10K whole-genome data in 1.5 hours, including QC, annotation, and association testing.
• An exome-wide significant association with CCDC62 was identified for rare variant aggregate analysis.
• SEQSpark demonstrated significant speed improvements, reducing computation time by up to 100-fold compared to other tools.

Conclusions:

• SEQSpark is a highly efficient and scalable tool for analyzing large-scale sequence data.
• The development of SEQSpark empowers large epidemiological studies to accelerate the discovery of genetic variants underlying complex traits.
• This tool facilitates faster and more comprehensive genetic association studies.