Identification of transcriptome-wide, nut weight-associated SNPs in Castanea crenata

Abstract

Nut weight is one of the most important traits that can affect a chestnut grower’s returns. Due to the long juvenile phase of chestnut trees, the selection of desired characteristics at early developmental stages represents a major challenge for chestnut breeding. In this study, we identified single nucleotide polymorphisms (SNPs) in transcriptomic regions, which were significantly associated with nut weight in chestnuts (Castanea crenata), using a genome-wide association study (GWAS). RNA-sequencing (RNA-seq) data were generated from large and small nut-bearing trees, using an Illumina HiSeq. 2000 system, and 3,271,142 SNPs were identified. A total of 21 putative SNPs were significantly associated with chestnut weight (false discovery rate [FDR] < 10−5), based on further analyses. We also applied five machine learning (ML) algorithms, support vector machine (SVM), C5.0, k-nearest neighbour (k-NN), partial least squares (PLS), and random forest (RF), using the 21 SNPs to predict the nut weights of a second population. The average accuracy of the ML algorithms for the prediction of chestnut weights was greater than 68%. Taken together, we suggest that these SNPs have the potential to be used during marker-assisted selection to facilitate the breeding of large chestnut-bearing varieties.

Single Nucleotide Polymorphisms as Practical Molecular Tools to Support European Chestnut Agrobiodiversity Management

“…Recent studies in Japanese and European chestnuts evidenced the need for molecular markers more suitable for cost effective and easy-to-use applications [ 13 , 23 ]. In molecular breeding, as well as in other field applications, the request of a high-throughput workflow from DNA isolation to data elaboration has already oriented researchers towards analyses based on SNP markers [ 11 , 22 , 31 ]. Nonetheless, until now, SNPs were mined within C. crenata and interspecific C. crenata × C. sativa breeding progenies, while information about C. sativa -specific SNPs was lacking.…”

“…However, the SNPs originally mined in C. crenata were poorly transferable to our genetic pool, as only four of 64 (6.25%) were polymorphic within tested European chestnuts. The molecular tools proposed in the present paper are SNP based, as suggested [ 11 , 13 , 23 , 31 , 32 ] and, as added value, each KASP assay can be analyzed independently from the others. In fact, KASP can be considered perfectly scalable, as a single plate can include from one to 380 samples and, thanks to the standardized cycling protocol, more than a SNP assay can be included if needed [ 28 ].…”

Machine learning, transcriptome, and genotyping chip analyses provide insights into SNP markers identifying flower color in Platycodon grandiflorus

“…Next-generation sequencing technologies have enabled large-scale genome-wide genotyping for heterogeneous phenotypes, which helped in precise genome selection associated with specific phenotypes. Further, the reduced representation of genome-wide genotyping is transcriptome-wide association studies, which leverage the project cost for molecular breeding studies in the model and non-model plants 8 – 10 . These massive genotyping efforts have recently been subject to machine learning (ML) methods to predict SNP associations with specific traits.…”

“…ML facilitates pattern recognition of large biological datasets since ML algorithms are used widely in various biological fields, such as molecular marker identification, coding region recognition, pathway gene recognition, protein–protein interaction determination, and metabolic network detection 12 . For instance, ML models have been constructed for genomic selection in wheat 13 , root genotype classification 14 , nut-size prediction in Castanea crenata 10 , and polyploidy associated SNPs identification in plants 15 .…”

Innovative transcriptome‐based genotyping highlights environmentally responsive genes for phenology, growth and yield in a non‐model grain legume

“…As the number of SNPs in coding regions is considerably lower than in noncoding regions, there was uncertainty about how useful the RNAseq variants would be for GWAS, particularly in crops with high linkage disequilibrium (LD) (Scheben, Batley, & Edwards, 2017). Nevertheless, performed studies demonstrated that transcriptome-based markers (SNPs and/or expression levels) might be also useful for prediction of complex traits, such as seed glucosinolate content in rapeseed (Lu et al, 2014); plant height, stem strength and spike architecture in wheat (Miller et al, 2016;Wang et al, 2017); nut weight in chestnuts (Kang et al, 2019); as well as plant height, flowering time and yield in maize (Azodi, Pardo, VanBuren, de los Campos, & Shiu, 2020). Recently, an innovative method of wholeexome genotyping and gene expression quantification based on massive analysis of cDNA ends (MACE) has been developed (Zawada et al, 2014).…”