Genome-wide approaches for the identification of markers and genes associated with sugarcane yellow leaf virus resistance

Abstract

The breeding of sugarcane, a leading sugar and energy crop, is complicated by the extremely complex sugarcane genome, which burdens research in the area and delays the development of new cultivars. One of the main viral diseases that affect this crop is sugarcane yellow leaf (SCYL), which is caused by the sugarcane yellow leaf virus (SCYLV). The most common symptom of SCYL is the yellowing of leaf midribs and blades, but asymptomatic cases are frequent. Regardless of the manifestation of SCYL, infection by SCYLV can lead to substantial yield losses, making resistance to this virus highly relevant to sugarcane breeding. However, the genetic basis of this trait has not been widely explored or explained. In this context, genome-wide association studies (GWASs) have arisen as promising tools for the identification of molecular markers associated with SCYLV resistance that can be employed in marker-assisted selection. In the present work, we performed a GWAS on sugarcane using codominant markers and genotypes of interest for breeding. A panel of 97 sugarcane genotypes inoculated with SCYLV was analyzed for SCYL symptom severity, and viral titer was estimated by reverse transcription quantitative PCR (RT-qPCR). A genotyping-by-sequencing (GBS) library was constructed for 94 individuals of this population, enabling the identification of 38,710 SNPs and 32,178 indels with information on allele proportion (AP) and position on the Saccharum spontaneum genome. For association analyses, several combinations of population structure and kinship were tested to reduce model inflation, and diverse marker-trait association mixed models were employed. We identified 35 markers significantly associated with SCYL symptom severity and 22 markers strongly associated with SCYLV titer that can be applied in breeding programs upon validation. By aligning the sequences flanking these markers with their coding sequences in several plant species, we annotated the functions of 7 genes. The possible involvement of these candidates in the response to SCYLV infection is discussed.

A joint learning approach for genomic prediction in polyploid grasses

“…This strategy of FS has already demonstrated promising results across the literature 27,[107][108][109][110] . Subsetting datasets through principles of FS is a common strategy in data science for datasets with a large number of variables 111 .…”

“…For SU SNP data, an SU pseudoreference genome was used for SNP calling 55 , different from the approach applied to FGs, where SNPs were estimated using genomic references from closely related species. Using Sorghum bicolor as a genomic reference for SU, a close relative, also decreases the quantity of reliable markers 27,110 and, consecutively, the chances of finding SNPs surrounding QTL regions. For this reason, we believe that the increases in accuracy in FG results were more modest than those in SU.…”

Multiomic investigation of sugarcane mosaic virus resistance in sugarcane

“…Interestingly, one marker identified through FS (Chr2A_103190628) was found to be close (1.7 kb) to this region. Linkage disequilibrium is high in sugarcane, persisting for up to 2-3.5 Mb (Yang et al ., 2019b; Pimenta et al ., 2021). Thus, it is possible that this marker is linked to Sspon.02G0027920-1A, the S. spontaneum gene syntenic to the auxin-binding protein gene at Scmv2 (Ding et al ., 2012).…”

“…These hybrids have large (D’Hont et al ., 1998), highly polyploid (D’Hont and Glaszmann, 2001), aneuploid (Sforça et al ., 2019) and duplicated (Aono et al, 2021) genomes that hinder sugarcane breeding research. Additionally, studies suggest that the majority of sugarcane traits are controlled by many small-effect loci (Gouy et al ., 2015; Fickett et al ., 2019; Pimenta et al ., 2021). However, given the existence of Scmv1 and Scmv2 in maize, it is odd that no major loci controlling SCMV resistance in sugarcane have been identified.…”

A divide-and-conquer approach for genomic prediction in rubber tree using machine learning

“…In this context, FS techniques may be an alternative strategy for building a predictive model, considering that not all markers are related to a specific phenotype (Yin et al, 2019) and that the quantity required for this task directly depends on the complexity and genetic architecture of the traits used (Liu et al, 2018). Therefore, like Bermingham et al (2015), Bellot et al (2018), Li et al (2018), Inácio & Alves (2019), Aono et al (2020), Ramzan et al (2020), Luo et al (2021), and Pimenta et al (2021), we decided to test the prediction improvements by using an FS technique to enhance network performances.…”