Accuracy of Four Heuristics for the Full Sibship Reconstruction Problem in the Presence of Genotype Errors

The full sibship reconstruction (FSR) problem is the problem of inferring all groups of full siblings from a given population sample using genetic marker data without parental information. The FSR problem remains a significant challenge for computational biology, since an exact solution for the problem has not been found. The new algorithm, named SIMPSON-assisted Descending Ratio (SDR), is devised combining a new Simpson index based O(n2) algorithm (MS2) and the existing Descending Ratio (DR) algorithm. The SDR algorithm outperforms the SIMPSON, MS2, and DR algorithms in accuracy and robustness when tested on a variety of sample family structures. The accuracy error is measured as the percentage of incorrectly assigned individuals. The robustness of the FSR algorithms is assessed by simulating a 2% mutation rate per locus (a 1% rate per allele).