optiCall: a robust genotype-calling algorithm for rare, low-frequency and common variants.

Area of Science:

• Genetics and Genomics
• Bioinformatics
• Computational Biology

Background:

• Current genotype calling algorithms for microarrays utilize either SNP-wise or sample-wise approaches.
• These methods face challenges in accurately calling genotypes, particularly for rare and low-frequency variants.
• Existing algorithms can lead to data loss due to false quality control failures and misclassification of rare variants.

Purpose of the Study:

• To develop and evaluate optiCall, a novel genotype calling algorithm for the Illumina platform.
• To improve the accuracy of genotype calling for rare, low-frequency, and common variants.
• To reduce the proportion of SNPs lost to downstream analysis compared to existing algorithms.

Main Methods:

• Developed optiCall, a genotype calling algorithm integrating both SNP-wise and sample-wise approaches.
• Implemented optiCall in C++ for UNIX operating systems.
• Evaluated optiCall using genotype data from 4537 individuals from the 1958 British Birth Cohort on the Immunochip.

Main Results:

• optiCall achieved a data loss of only 1.38% due to false quality control failures and misclassification of rare variants.
• This is a significant improvement compared to existing algorithms: Illuminus (3.87%), GenoSNP (7.85%), and GenCall (4.09%).
• optiCall accurately captures rare variants and correctly accounts for SNPs with shifted probe intensity clouds.

Conclusions:

• optiCall offers a more accurate and efficient solution for genotype calling on the Illumina platform.
• The algorithm effectively minimizes data loss, particularly for rare variants, enhancing the reliability of downstream genetic analyses.
• optiCall is available for download, facilitating its adoption in genetic research.