Fig 1.
Overview of the method for pairwise kinship analysis proposed in this study.
ICS is the index of chromosome sharing. C-Th and L-Th indicate the most appropriate thresholds for collateral relatives and for lineal relatives, respectively.
Fig 2.
The expected ICS variability for each relationship.
Boxplots indicate the variability from the computationally generated genotypes. Triangular plots indicate the values from half of the actual samples. (A) ICSC variability for collateral relatives using only IBS segments longer than 4 cM and (B) ICSL variability for lineal relatives using only IBS segments longer than 3 cM.
Fig 3.
Q-Q plots for the ICSC values calculated by the genotypes of computationally generated C-5.
These plots are assumed as (A) the normal distribution, (B) the truncated-normal distribution, and (C) the log-normal distribution.
Table 1.
Three quantiles of LR values for the actual sample pairs obtained by comparing the true relationship with the UN.
Fig 4.
The range of ICS values in each Hummel’s predicate for kinship determination.
ICS values for (A) collateral relatives and (B) lineal relatives. Solid lines indicate the estimated log-normal distributions of the ICSC and ICSL values. The colors in the bar under the distributions indicate the range of the ICSC and ICSL values that can be obtained from the posterior probabilities corresponding to each Hummel’s predicate for each relationship. The plots indicate the ICSC and ICSL values calculated by the rest of the actual sample pairs. All actual L-1 pairs were almost equal to the maximum value of ICSL (i.e., 3662.522) and are indicated in red. * The distribution of UN is partially displayed because the density of the estimated distribution of UN is high.
Table 2.
Percentages of the posterior probabilities ≥ 0.998, 0.99, 0.95, and 0.90 for the distributions of each relationship.
Table 3.
The number of samples categorized by the range of posterior probabilities based on Hummel’s predicates.
Fig 5.
Pedigree consisted of twelve people.