Insights into gene conversion and crossing-over processes from long-read sequencing of human, chimpanzee and gorilla testes and sperm

Abstract

AbstractHomologous recombination rearranges genetic information during meiosis to generate new combinations of variants. Recombination also causes new mutations, affects the GC content of the genome and reduces selective interference. Here, we use HiFi long-read sequencing to directly detect crossover and gene conversion events from switches between the two haplotypes along single HiFi-reads from testis tissue of humans, chimpanzees and gorillas as well as human sperm samples. Furthermore, based on DNA methylation calls, we classify the cellular origin of reads to either somatic or germline cells in the testis tissue. We identify 1692 crossovers and 1032 gene conversions in nine samples and investigate their chromosomal distribution. Crossovers are more telomeric and correlate better with recombination maps than gene conversions. We show a strong concordance between a human double-strand break map and the human samples, but not for the other species, supporting different PRDM9-programmed double-strand break loci. We estimate the average gene conversion tract lengths to be similar and very short in all three species (means 40-100 bp, fitted well by a geometric distribution) and that 95-98% of non-crossover events do not involve tracts intersecting with polymorphism and are therefore not detectable. Finally, we detect a GC bias in the gene conversion of both single and multiple SNVs and show that the GC-biased gene conversion affects SNVs flanking crossover events. This implies that gene conversion events associated with crossover events are much longer (estimated above 500 bp) than those associated with non-crossover events. Highly accurate long-read sequencing combined with the classification of reads to specific cell types provides a new, powerful way to make individual, detailed maps of gene conversion and crossovers for any species.