Estimation of single-generation sibling relationships based on DNA markers

An algorithm for infering joint sibling relationships for a single generation sample of organisms based on DNA marker data is presented. The technique is based primarily on the exclusion principle, i.e., the process of systematically eliminating hypothetical kinship types between individuals in the basis of DNA incompatibility. The result is an enumeration of all genetically compatible sibling groupings, which is maximal in the sense that no other individual can be added to the sibling group. The technique is computationally feasible for large samples given enough variablility in the DNA markers used. From this, we construct a partition of the individuals into sibling groups based on a score. In addition, we derive some statistical properties of sibling compatibility events as well as various types of error probabilities associated with sibling identification. The techniques are illustrated using fisheries data.