Refined genetic maps reveal sexual dimorphism in human meiotic recombination at multiple scales

Abstract

In humans, males have lower recombination rates than females over the majority of the genome, but the opposite is usually true near the telomeres. These broad-scale differences have been known for decades, yet little is known about differences at the fine scale. By combining data sets, we have collected recombination events from over 100,000 meioses and have constructed sex-specific genetic maps at a previously unachievable resolution. Here we show that, although a substantial fraction of the genome shows some degree of sexually dimorphic recombination, the vast majority of hotspots are shared between the sexes, with only a small number of putative sex-specific hotspots. Wavelet analysis indicates that most of the differences can be attributed to the fine scale, and that variation in rate between the sexes can mostly be explained by differences in hotspot magnitude, rather than location. Nonetheless, known recombination-associated genomic features, such as THE1B repeat elements, show systematic differences between the sexes.

Figure 1. Improved resolution of the refined genetic maps.

(a) Localization of recombination events before and after refinement by the MCMC procedure. (b) Squared Pearson correlation as a function of scale between the sex-averaged maps and the HapMap map. (c) Mean recombination rate around hotspots defined in the LD-based HapMap map before and after refinement. (d) Sex-specific recombination rates in the refined map around HapMap recombination hotspots. (e) Cumulative proportion of recombination in the genetic map versus the proportion of sequence.

Figure 2. Fine-scale differences in recombination rates between females and males.

(a) Fraction of the autosomal genome with female recombination rate higher than the male rate (red) and male rate higher than female (blue). (b) An example of sex-specific hotspot on chromosome 15, recombining at higher rate in males (blue) than females (red). The 10 kb dimorphic region is centred around an inter-SNP interval showing significant sex-difference in rate based on the 99% credible intervals (CI) computed in our MCMC mapping method.

Figure 3. Wavelet analysis of female and male recombination rates.

(a) Continuous wavelet transformation of recombination rates along a 2 Mb region of chromosome 10. Line plots represent the original signal to which the wavelet transform was applied, namely the female (red) and male (blue) log-transformed recombination rates, and the difference between the two (green). Scalograms represent the continuous wavelet transformation (CWT) coefficients for scales from 2 kb up to 512 kb. Colours indicate the magnitude of wavelet coefficients (blue=negative, yellow=positive) at each scale and location, with each level normalized to have equal variance. (b) Genome-wide power spectrum of the female and male recombination rates at scales from 2 to 16,384 kb. (c) Correlation between the detail coefficients of the DWT of female and male recombination rates as a function of scale. The colour of each bar indicates the P value of the correlation, with smaller values shown in darker red shades.

Figure 4. Genomic features associated with the recombination rate.

Wavelet-based linear model of scale-specific genomic correlates to the recombination rate. Table shows the marginal significance (−log10 P-value two-sided t-test) for the linear model analyses of wavelet detail coefficients. Colours indicate the direction of the relationship (red=positive; blue=negative) with intensity proportional to significance. Linear regression analysis was performed on the log-transformed (a) female recombination rate, (b) male recombination rate and (c) on the sex difference between the log-transformed rates (female–male).

Figure 5. Sex-specific recombination rates around transcription start sites.

Average female (red) and male (blue) recombination rates in 1 kb bins around the transcription start sites (TSS) of a subset of 15,239 GENCODE genes randomly selected to be spaced from each other by 5 kb or more. The 95% confidence interval around the mean estimates of recombination rates are shown in shaded colours. (a) All TSS and (b) TSS partitioned according to the presence/absence of a PRDM9 degenerate 13-mer motif within 5 kb.

Figure 6. Relationship between H3K4 trimethylation and sex-specific recombination rates marks.

(a) Average female and male recombination rates around the middle point of H3K4me3 peaks from testis sample. Recombination rates around H3K4me3 marks partitioned into (b) those that overlap with predicted degenerate 13-mer motif and those that do not overlap any, and (c) those exclusive to testis sample, shared between testis sample and 41 ENCODE cell lines and found in ENCODE cell lines but not in testis. The 95% confidence interval around the mean estimates of recombination rates are shown in shaded colours.