Y-chromosome based evidence for pre-neolithic origin of the genetically homogeneous but diverse Sardinian population: inference for association scans

Abstract

The island of Sardinia shows a unique high incidence of several autoimmune diseases with multifactorial inheritance, particularly type 1 diabetes and multiple sclerosis. The prior knowledge of the genetic structure of this population is fundamental to establish the optimal design for association studies in these diseases. Previous work suggested that the Sardinians are a relatively homogenous population, but some reports were contradictory and data were largely based on variants subject to selection. For an unbiased assessment of genetic structure, we studied a combination of neutral Y-chromosome variants, 21 biallelic and 8 short tandem repeats (STRs) in 930 Sardinian males. We found a high degree of interindividual variation but a homogenous distribution of the detected variability in samples from three separate regions of the island. One haplogroup, I-M26, is rare or absent outside Sardinia and is very common (0.37 frequency) throughout the island, consistent with a founder effect. A Bayesian full likelihood analysis (BATWING) indicated that the time from the most recent common ancestor (TMRCA) of I-M26, was 21.0 (16.0-25.5) thousand years ago (KYA) and that the population began to expand 14.0 (7.8-22.0) KYA. These results suggest a largely pre-Neolithic settlement of the island with little subsequent gene flow from outside populations. Consequently, Sardinia is an especially attractive venue for case-control genome wide association scans in common multifactorial diseases. Concomitantly, the high degree of interindividual variation in the current population facilitates fine mapping efforts to pinpoint the aetiologic polymorphisms.

Figure 1. Map of Mediterranean showing the location of Sardinia.

Expanded section shows the island of Sardinia with the sample regions highlighted

Figure 2. UPGMA consensus tree correlating European populations constructed using 1000 F_ST pairwise distances.

Numbers before forks indicate percentage of times that each group appeared in the input trees

Figure 3. UPGMA consensus tree correlating European populations constructed using 1000 R_ST pairwise distances.

Numbers before forks indicate percentage of times that each group appeared in the input trees