Abstract
Long persistence of allelic lineages, prevalence of nonsynonymous over synonymous substitutions in the peptide-binding region (PBR), and deviation from neutrality of the expected gene identity parameter F all indicate indirectly that balancing selection is operating at functional major histocompatibility complex (MHC) loci. Direct demonstrations of the existence of balancing selection at MHC loci are, however, either lacking or not fully convincing. To define the conditions under which balancing selection could be demonstrated, we estimated its intensity from the mean number of nonsynonymous substitutions, KB, at the PBR and the mutation rate mu. We compared the five available methods for estimating KB by computer simulation and chose the most reliable ones for estimation of selection intensity. For the human MHC, the selection coefficients of the HLA-A, -B, -C, -DRB1, -DQB1, -DQA1, and -DPB1 loci are 0.015, 0.042, 0.0026, 0.019, 0.0085, 0.0028, and 0.0007, respectively. This low selection intensity places severe restrictions on the possibility of measuring selection directly in vertebrate populations.