Skip to main content
. 2016 Oct 10;11(10):e0163973. doi: 10.1371/journal.pone.0163973

Fig 3. Workflow.

Fig 3

Each individual's genotype (g1… gK) is interpreted as possibly ambiguous pairs of haplotypes (h1… hH), whose frequencies are initialized evenly. The EM algorithm then alternates between the E and M steps until convergence. The results are haplotype pair assignments for each individual and frequencies estimates for each haplotype.