Extended Data Fig. 2. Relative effective sample size for estimating direct genetic effects when imputing a missing parent’s genotype from a parent–offspring pair using unphased data.

We derive that the relative effective sample is $\frac{1-3f\left(1-f\right)}{2-2f\left(1-f\right)}$ compared to using fully observed parental genotypes (Supplementary Note Section 5.1). There is a penalty for heterozygosity when using unphased data since the allele that is shared with the observed parent cannot be determined when both parent and offspring are heterozygous. In contrast, the relative efficiency when using phased data is 1/2, independent of allele frequency.