Genotype error biases trio-based estimates of haplotype phase accuracy

Area of Science:

• Genetics
• Bioinformatics
• Statistical genomics

Background:

• Statistical phasing of unphased genotype data is crucial for genetic research.
• Accuracy of statistical phasing is typically assessed using the switch error rate in parent-offspring trios.
• Genotype errors can introduce bias into phasing accuracy metrics.

Purpose of the Study:

• To develop methods for estimating genotype error rates from parent-offspring trios.
• To quantify the bias in switch error rates caused by genotype errors.
• To propose an improved definition of phase error for more accurate assessment.

Main Methods:

• Applied statistical methods to estimate genotype error rates using parent-offspring trios.
• Developed a method to estimate bias in observed switch error rates due to genotype errors.
• Analyzed large datasets from UK Biobank and TOPMed, including trio data.

Main Results:

• Genotype errors were shown to inflate the observed switch error rate.
• The relative bias in switch error rate increases with sample size.
• For UK Biobank White British trios, the observed switch error rate was 2.4 times higher than the true rate.
• An alternate phase error definition estimated an average distance of 64 megabases between phase errors in UK Biobank White British individuals.

Conclusions:

• Genotype error is a significant confounder in assessing statistical phasing accuracy.
• The proposed methods provide a more accurate estimation of phasing errors.
• Improved phase error metrics are essential for large-scale genetic datasets and accurate haplotype inference.