Model-P: a basecalling method for resequencing microarrays of diploid samples

Basecalling is a critical step of the analysis of DNA resequencing microarray data for single nucleotide polymorphism discovery and genotyping. For microarrays hybridized with DNA derived from diploid organisms, basecalling with high accuracy at high call rates is a challenging task. Current methods sometimes do not produce satisfactory results.

We explored using physical models based on the sequences of the probe and the target to predict feature intensities in resequencing microarrays. Based on these intensity-predicting models, a new basecalling method (Model-P), which takes into consideration the expected feature intensities for different potential genotypes, was developed. Model-P is shown to have better performance at high call rates compared with ABACUS, the current state-of-the-art method, on a test dataset and on relatively AT-rich regions.

Model-P is available upon request.