SNPCaller

SNPCaller improves genotype calling accuracy from raw intensity data for single nucleotide polymorphism (SNP) markers in disomic and trisomic datasets by applying model-based clustering and incorporating pedigree information.

Key Features:

• Model-Based Clustering: Employs mixture models, including Gaussian and beta-mixture models, and selects models based on prior knowledge of measurement distributions to distinguish genotype clusters.
• Incorporation of Pedigree Information: Integrates pedigree information to inform genotype calls by considering familial relationships and shared genetic backgrounds.
• Trisomic Genotype Calling: Calls genotypes for trisomic individuals by accounting for the additional chromosome present in conditions such as Down syndrome.
• Modified K-means Method: Updates all family members' genotype calls simultaneously using pedigree information to refine clustering.
• Likelihood-Based Approach: Combines mixture models with pedigree data in a probabilistic framework to compute likelihoods for genotype predictions.

Scientific Applications:

• Genetic linkage and association studies: Provides improved genotype calls for family-based analyses in linkage and association studies.
• Clinical genetics: Supports genotyping in studies of trisomic conditions such as Down syndrome to aid diagnosis and characterization.
• Population genetics: Enhances SNP genotype accuracy for population-scale analyses requiring reliable genotype calls.

Methodology:

Performance is compared against existing clustering techniques using simulation studies, and methods are validated on real datasets generated by platforms such as Illumina.