Measuring the invisible - The sequences causal of genome size differences in eyebrights (Euphrasia) revealed by k-mers

Abstract

Genome size variation within plant (and other) taxa may be due to presence/absence variation in low-copy sequences or copy number variation in genomic repeats of various frequency classes. However, identifying the sequences underpinning genome size variation has been challenging because genome assemblies commonly contain collapsed representations of repetitive sequences and because genome skimming studies miss low-copy number sequences. Here, we take a novel approach based on k-mers, short sub-sequences of equal length k, generated from whole genome sequencing data of diploid eyebrights (Euphrasia), a group of plants which have considerable genome size variation within a ploidy level. We compare k-mer inventories within and between closely related species, and quantify the contribution of different copy number classes to genome size differences. We further assign high-copy number k-mers to specific repeat types as retrieved from the RepeatExplorer2 pipeline. We find complex patterns of k-mer differences between samples. While all copy number classes contributed to genome size variation, the largest contribution came from repeats with 1000-10,000 genomic copies including the 45S rDNA satellite DNA and, unexpectedly, a repeat associated with an Angela transposable element. We also find size differences in the low-copy number class, likely indicating differences in gene space between our samples. In this study, we demonstrate that it is possible to pinpoint the sequences causing genome size variation within species without use of a reference genome. Such sequences can serve as targets for future cytogenetic studies. We also show that studies of genome size variation should go beyond repeats and consider the whole genome. To allow future work with other taxonomic groups, we share our analysis pipeline, which is straightforward to run, relying largely on standard GNU command line tools.