Estimating the Rate of Gene Conversion on Human Chromosome 21

There is a growing recognition that gene conversion can be an important factor in shaping fine-scale patterns of linkage disequilibrium in the human genome. We devised simple multilocus summary statistics for estimating gene-conversion rates from genomewide polymorphism data sets. In addition to being computationally feasible for very large data sets, these summaries were designed to yield robust estimates of gene-conversion rates in the presence of variation in crossing-over rates. Using our summaries, we analyzed 21,840 biallelic single-nucleotide polymorphisms (SNPs) on human chromosome 21. Our results indicate that models including both crossing over and gene conversion fit the overall short-range data (0.5 kb) of chromosome 21 much better than do models including crossing over alone. The estimated ratio of gene-conversion rate to crossing-over rate has a range of 1.6.9.4, depending on the assumed conversion tract length (in the range of 500.50 bp). Removal of the 5,696 SNPs that occur in known mutational hotspots (CpG sites) did not significantly change our conclusions, suggesting that recurrent mutations alone cannot explain our data.