A Cheap and Efficient Approach to Identify Sex-Linked Genes in Non-Model Organisms using NGS Data

Most animals and ~6% of flowering plants have separated sexes but the underlying genetic bases are poorly known in most cases. The advance of next generation sequencing technologies provides an unprecedented opportunity to study sex determination and test the presence of sex chromosomes in “non-model” plant and animal species. One very promising approach relies upon RNAseq data from a cross (parents and progeny) and was successfully applied to the dioecious plant Silene latifolia.

However, identifying genes located on sex chromosomes from RNAseq data reliably is not a simple task and this was done rather empirically in previous work. Here we propose a new pipeline that relies on genotypes from a genotyper designed for species without reference genome. It can also deal with differential allele expression, which is important for detecting sex chromosomes as the Y copy usually shows reduced expression. We then introduce a new method based on a model of allele transmission from parents to progeny to compute the probability of a contig to be sex-linked (both X and Y copies), X hemizygous (X-linked only) and autosomal, which uses an expectation maximisation (EM) algorithm. Our pipeline was tested on a set of experimentally known autosomal and sex-linked genes in S. latifolia and gave better results than previous work. It was also tested using simulations and will be available in a Galaxy workflow.

Our pipeline will be very useful to detect sex chromosomes and to identify sex-determining genes in a number of non-model plants and animals with separated sexes.