Exploring the impact of sequence context on errors in SNP genotype calling with whole genome sequencing data using...

Area of Science:

• Genomics
• Bioinformatics
• Animal Genetics

Background:

• Variant calling is crucial for whole genome sequencing (WGS) data analysis but is error-prone.
• Incorrect single nucleotide polymorphism (SNP) calls can impact downstream genomic analyses.
• Identifying patterns in variant calling errors is essential for improving data quality.

Purpose of the Study:

• To investigate the association between incorrect SNP calls and variant quality metrics.
• To identify systematic patterns in SNP calling errors using nucleotide context.
• To develop methods for detecting and potentially correcting erroneous SNP calls in WGS data.

Main Methods:

• Compared SNP genotypes from WGS (Illumina NovaSeq 6000) with a genotyping microarray (EuroGMD50K) in Holstein-Friesian cows.
• Defined correct (666,333 SNPs) and incorrect (4,557 SNPs) SNP sets.
• Employed an autoencoder, one-class support vector machine, and isolation forest algorithms to identify systematic errors.

Main Results:

• Approximately 59.53% of incorrect SNPs exhibited systematic patterns, while the rest were random errors.
• The sequence context 'CGC' was frequently associated with miscalled 'C' variants.
• An incorrect 'T' instead of 'A' call was linked to a downstream 'T' nucleotide.

Conclusions:

• Systematic errors in SNP calling are prevalent and associated with specific sequence contexts and nucleotide labeling patterns.
• These findings provide insights into the sources of variant calling errors in WGS data.
• Improved understanding of error patterns can lead to more accurate genomic analyses in cattle and other species.