Are human hemoglobin variants distributed randomly among the positions?

Summary

The distribution of human hemoglobin variants has previously been studied by Vogel (1972) who concluded that the distribution was random although no statistical analysis was presented. This work points out that there are four biases in the data, one in the manner in which the number of variants is counted, another in the method by which they are detected and which favors charge changes, a third in the fact that for a few codons the same amino acid replacement may be brought about by two or three single nucleotide replacements, and a fourth in the non-random sampling procedure which favors variants producing clinical symptoms. Nevertheless, the distribution of beta hemoglobin variants is confirmed to be random as Vogel suggests. The alpha hemoglobin variants are distinctly non-randomly distributed, the best fit requiring that 69 of the alpha positions be considered invariable. The above biases could account for this result but other considerations combine to suggest the following: 1, about half of all alterations of alpha hemoglobin will not survive to sampling whereas nearly all beta variants can; 2, deleterious mutants that survive to sampling but are destined to be eliminated by selection are more likely to be observed in beta than in alpha hemoglobin; and 3, mutations destined to go to fixation are more likely to occur in beta than in alpha hemoglobin.