HipSTR was specifically developed to deal with these errors in the hopes of obtaining more robust STR genotypes. In particular, it accomplishes this by:
Learning locus-specific PCR stutter models using an EM algorithm
Mining candidate STR alleles from population-scale sequencing data
Employing a specialized hidden Markov model to align reads to candidate alleles while accounting for STR artifacts
Utilizing phased SNP haplotypes to genotype and phase STRs
