Disorders can often have genotype information from only one parent of the affected individuals. As a common practice, these trios are simply discarded [42] though this can result in considerable loss of information and bias to the association study [20], [38]. Several studies have been proposed to allow TDT to handle missing parental genotypic information [20], [22], [43]–[50]. Within these studies, the missing parental genotypes are mostly reconstructed based on the assumption that they are missing completely at random and do not depend on the genotypes themselves [51]. However, this assumption may not hold true and the probability that a genotype is missing may rely on the unobserved alleles [38], [52], [53]. Furthermore, these approaches are not designed for pedigrees with missing genotypes on the proband when both linkage and association data are available.
