Plink really does a great job when it calculate LD and even for multiple haplotypes/phasings, LDs are provided. However, the posterior probability were lost, right? How to choose the best one when it happened like this situation?

> --ld rs998731 rs77356712: Multiple haplotype phasing solutions; sample size, HWE, or random mating assumption may be violated.
> 
> HWE exact test p-values
> -----------------------    rs998731: 0.549867    rs77356712: 1
> 
> Solution #1:    R-sq = 0.00979996     D' = 1
> 
>    Haplotype     Frequency    Expectation under LE    ---------    
> ---------    --------------------
>           TA      0                       0.004948
>           CA      0.010101                0.005153
>           TG      0.489899                0.484951
>           CG      0.5                     0.504948
> 
>    In phase alleles are TG/CA
> 
> Solution #2:    R-sq = 0.0106249      D' = 1
> 
>    Haplotype     Frequency    Expectation under LE    ---------    
> ---------    --------------------
>           TA      0.010101                0.004948
>           CA      0                       0.005153
>           TG      0.479798                0.484951
>           CG      0.510101                0.504948
> 
>    In phase alleles are TA/CG