THE POWER OF AN IBS-BASED METHOD TO INFER RELATIONSHIPS BETWEEN PAIRS OF INDIVIDUALS Silvano Presciuttini 1, Chiara Toni 1, Simonetta Verdiani 2, Lucia Casarino 2, Isabella Spinetti 2, Francesco De Stefano 2, Ranieri Domenici 1 1 Dept. of Biomedicine, Univ. of Pisa, Italy 2 Dept. Of Legal Medicine, Univ. of Genova, Italy

OUTLINE OF THE IBS METHOD Consider two subjects that we want to test for relationship. They share 0, 1, or 2 alleles identical by state (IBS) at any given locus. The probabilities z 0, z 1, and z 2 of the three possible outcomes depend on the genetic variability of the locus and on the degree of their relationship. For parent-child pairs, it is easy to show that, z 0 = 0, z 1 = H and z 2 = 1-H, where H is the locus heterozygosity. In the case of other relationships, the dependence of z i on H is more complex.

Probabilities of sharing zero alleles (z 0 ) at 19 loci as a function of H for three common relationships Locus heterozygosity (H) NR: non-relatives, 2D: all 2nd degree, FS: full sibs. Lines: third-order polynomial regression curves

Probabilities of sharing one alleles (z 1 ) at 19 loci as a function of H for three common relationships NR: non-relatives, 2D: all 2nd degree, FS: full sibs. Lines: third-order polynomial regression curves Locus heterozygosity (H)

Probabilities of sharing both alleles (z 2 ) at 19 loci as a function of H for three common relationships Locus heterozygosity (H) NR: non-relatives, 2D: all 2nd degree, FS: full sibs. Lines: third-order polynomial regression curves

Full sibs z H – H H 3 z – H H 2 – H 3 z H – H H 3 Second degree z H – H H 3 z – H H 2 – H 3 z H – H H 3 Non-relatives z H – H H 3 z – H H 2 – H 3 z H – H H 3 Equations relating H (locus heterozygosity) to z i

Table of z i probabilities for four common relationships and 18 commonly used markers

Contrasting likelihood ratios computed by the exact method and by the IBS method (1) 1.Likelihood ratios (log to base 10) that 102 true parent-child pairs are parent-child pairs rather than non- relatives, conditional on their genotypes at several loci, calculated by the exact method (X axis) and by the IBS method (Y axis).

Contrasting likelihood ratios computed by the exact method and by the IBS method (2) 2.Likelihood ratios (log to base 10) that 80 true full-sib pairs are siblings rather than non-relatives, conditional on their genotypes at 13 loci, calculated by the exact method (X axis) and by the IBS method (Y axis).

An example of the calculations needed for a given full-sib pair (1) chosen among 7 possible alternatives; (2) calculated from sample data; (3) FS rather than NR, computed using exact formulas; (4) FS rather than NR, computed from tabulated values

CONCLUSIONS The statistical power of the IBS method to infer relationships between individuals is not much lower than that of the exact method. The IBS method may be conveniently used as a preliminary approach. It can be applied by anybody using a desk calculator or a spreadsheet. In certain circumstances, the results of the IBS method may even be accepted without further analyses, since the LRs are highly correlated with those calculated by the exact method. Of course, the exact method should always follow IBS analysis when the results are critical to living human subjects.