Allele expression biases in mixed-ploid sugarcane accessions

Abstract

Allele-specific expression (ASE) represents differences in the magnitude of expression between alleles of the same gene. This is not straightforward for polyploids, especially autopolyploids, as knowledge about the dose of each allele is required for accurate estimation of ASE. This is the case for the genomically complex Saccharum species, characterized by high levels of ploidy and aneuploidy. We used a Beta-Binomial model to test for allelic imbalance in Saccharum, with adaptations for mixed-ploid organisms. The hierarchical Beta-Binomial model was used to test if allele expression followed the expectation based on genomic allele dosage. The highest frequencies of ASE occurred in sugarcane hybrids, suggesting a possible influence of interspecific hybridization in these genotypes. For all accessions, ASEGs were less frequent than those with balanced allelic expression. These genes were related to a broad range of processes, mostly associated with general metabolism, organelles, responses to stress and responses to stimuli. In addition, the frequency of ASEGs in high-level functional terms was similar among the genotypes, with a few genes associated with more specific biological processes. We hypothesize that ASE in Saccharum is largely a genotype-specific phenomenon, as a large number of ASEGs were exclusive to individual accessions.

Limited allele-specific gene expression in highly polyploid sugarcane

“…A Bayesian hierarchical beta-binomial model ( Correr et al 2021 ) confirmed that most SNVs showed no evidence of ASE, with 19,154 to 56,956 heterozygous sites yielding significant tests ( Table 1 ; Supplemental Fig. S6 ).…”

“…Different statistical models and testing strategies, hard thresholds, and other ad hoc criteria can affect the results. In sugarcane, Correr et al (2021) also observed ASE for ∼40% of the sampled genes. However, this study relied on genotyping-by-sequencing of reduced representation libraries, a strategy that is prone to biases and genotyping errors ( Scheben et al 2017 ).…”

“…A beta-binomial model was adapted to investigate ASE in mixed- and high-ploidy organisms ( Correr et al 2021 ), and this study further modified this method to model uncertainty in estimating allele doses. This approach allowed nonhomogeneous sequencing depth profiles to be naturally handled by the model, with a corresponding adjustment of the null hypothesis for each polymorphism.…”

“…To test for ASE, the hierarchical beta-binomial model proposed by Correr et al (2021) was used with modifications. We first modelled y ik according to a binomial distribution, .…”

“…There is also evidence of conserved ASE between maize and rice orthologs, with accompanying evidence of positive selection ( Waters et al 2013 ). The joint availability of high-throughput RNA sequencing (RNA-seq) assays and methods for quantitative genotyping of single-nucleotide variants (SNVs) has allowed ASE to be studied in sugarcane ( Vilela et al 2017 ; Sforça et al 2019 ; Cai et al 2020 ; Correr et al 2021 ). However, these previous investigations focused on a small number of genes, rely on limited genomic sampling, or suffer from potential sources of bias in the genotype calls.…”