Sex-detector: A probabilistic approach to study sex chromosomes in non-model organisms

Abstract

Wepropose a probabilistic framework to infer autosomal and sex-linked genes fromRNA-seq data of a cross for any sex chromosome type (XY, ZW, and UV). Sex chromosomes (especially the non-recombining and repeat-dense Y, W, U, and V) are notoriously difficult to sequence. Strategies havebeendeveloped toobtainpartially assembled sexchromosomesequences.Mostofthemremain difficult to apply to numerous non-model organisms, either because they require a reference genome, or because they are designed for evolutionarily old systems. Sequencing a cross (parents and progeny) by RNA-seq to study the segregation of alleles and infer sexlinked genes is a cost-efficient strategy, which also provides expression level estimates. However, the lack of a proper statistical frameworkhas limited abroader applicationof this approach. Testsonempirical Silenedata showthatourmethodidentifies20-35% more sex-linked genes than existing pipelines,while making reliable inferences for downstream analyses. Approximately 12 individuals are needed for optimal results based on simulations. For species with an unknown sex-determination system, the method can assess the presence and type (XY vs. ZW) of sex chromosomes through amodel comparison strategy. Themethod is particularlywell optimized for sex chromosomes of young or intermediate age, which are expected in thousands of yet unstudied lineages. Any organisms, including non-model ones for which nothing is known a priori, that can be bred in the lab, are suitable for our method. SEX-DETector and its implementation in a Galaxy workflow are made freely available.