Statistical Approach to Relatedness Analysis in Large Collections of Genetic Profiles. An Application to a DNA-Registry of Fin Whales

The present study utilized data from an Icelandic DNA-registry of fin whales by searching for pairs of relatives with a statistical approach. Three types of relatedness were of interest, half-siblings, parent-offspring and first cousins. Detection of relatives was done by computing pairwise LOD scores for the individuals in the sample for each relatedness of interest. The corresponding p-values for each LOD score were estimated by comparing the original LOD scores with LOD scores of simulated unrelated individuals. Due to very high number of pairwise comparisons, adjustment for multiple testing was necessary. Two well known multiple adjusting methods were applied and compared, the Bonferroni correction and the FDR procedure. The FDR procedure was found to be more suitable for this analysis since the Bonferroni procedure was too conservative for such a high number of LOD scores. Eight pairs of relatives were detected within the sample. When information about age and age of maturity had been taken into account, three of those pairs were classified as a parent-offspring pair and five were classified as half-siblings. One of the detected parent-offspring pairs were a male fin whale and a foetus, also detected as a father-offspring pair in a previous study. Master i Statistikk MAMN-STAT STAT399