Difficulty in detecting rare variants is one of the problems in conventional genome-wide association studies (GWAS). The problem is closely related to the complex gene compositions comprising multiple alleles, such as haplotypes. Several single nucleotide polymorphism (SNP) set approaches have been proposed to solve this problem. These methods, however, have been rarely discussed in connection with haplotypes. In this study, we developed a novel SNP-set method named “RAINBOW” and applied the method to haplotype-based GWAS by regarding a haplotype block as a SNP-set. Combining haplotype block estimation and SNP-set GWAS, haplotype-based GWAS can be conducted without prior information of haplotypes. We prepared 100 datasets of simulated phenotypic data and real marker genotype data of Oryza sativa subsp. indica, and performed GWAS of the datasets. We compared the power of our method, the conventional single-SNP GWAS, the conventional haplotype-based GWAS, and the conventional SNP-set GWAS. Our proposed method was shown to be superior to these in three aspects: (1) controlling false positives; (2) in detecting causal variants without relying on the linkage disequilibrium if causal variants were genotyped in the dataset; and (3) it showed greater power than the other methods, i.e., it was able to detect causal variants that were not detected by the others, primarily when the causal variants were located very close to each other, and the directions of their effects were opposite. By using the SNP-set approach as in this study, we expect that detecting not only rare variants but also genes with complex mechanisms, such as genes with multiple causal variants, can be realized. RAINBOW was implemented as an R package named “RAINBOWR” and is available from CRAN (https://cran.r-project.org/web/packages/RAINBOWR/index.html) and GitHub (https://github.com/KosukeHamazaki/RAINBOWR). Author summary Detecting rare variants has been one of the most problematic problems in GWAS. Here, we proposed a novel SNP-set GWAS approach, which is superior in controlling false positives and detecting rare variants compared with conventional approaches, and implemented this method as an R package named “RAINBOWR” (Reliable Association INference By Optimizing Weights with R). In this article, we introduce the application of RAINBOW to haplotype-based GWAS by regarding a haplotype block as a SNP-set, which enables one to perform haplotype-based GWAS without prior haplotype information. We showed that the haplotype-based GWAS with the RAINBOW package succeeded in detecting causal variants with complex mechanisms that were not detected by any other conventional methods. RAINBOW also offers a fast single-SNP GWAS method. RAINBOW offers not only a SNP-set GWAS that can be applied to universal situations but also one that is faster with the restircted situations using linear kernel for constructing the Gram matrix of SNP-set of interest. We also used Rcpp (functions for using C++ in R) for the RAINBOW implementation to achieve faster computation. We believe that our package will lead to the detection of novel genes associated with biologically and agronomically essential traits.

High-throughput phenotyping serves as a framework to reduce chronological costs and accelerate breeding cycles. In this study, we developed models to estimate the phenotypes of biomass-related traits in soybean (Glycine max) using unmanned aerial vehicle (UAV) remote sensing and deep learning models. In 2018, a field experiment was conducted using 198 soybean germplasm accessions with known whole-genome sequences under 2 irrigation conditions: drought and control. We used a convolutional neural network (CNN) as a model to estimate the phenotypic values of 5 conventional biomass-related traits: dry weight, main stem length, numbers of nodes and branches, and plant height. We utilized manually measured phenotypes of conventional traits along with RGB images and digital surface models from UAV remote sensing to train our CNN models. The accuracy of the developed models was assessed through 10-fold cross-validation, which demonstrated their ability to accurately estimate the phenotypes of all conventional traits simultaneously. Deep learning enabled us to extract features that exhibited strong correlations with the output (i.e., phenotypes of the target traits) and accurately estimate the values of the features from the input data. We considered the extracted low-dimensional features as phenotypes in the latent space and attempted to annotate them based on the phenotypes of conventional traits. Furthermore, we validated whether these low-dimensional latent features were genetically controlled by assessing the accuracy of genomic predictions. The results revealed the potential utility of these low-dimensional latent features in actual breeding scenarios.

A genome‐wide association study (GWAS) needs to have a suitable population. The factors that affect a GWAS (e.g. population structure, sample size, and sequence analysis and field testing costs) need to be considered. Mixed populations containing subpopulations of different genetic backgrounds may be suitable populations. We conducted simulation experiments to see if a population with high genetic diversity, such as a diversity panel, should be added to a target population, especially when the target population harbors small genetic diversity. The target population was 112 accessions of Oryza sativa L. subsp. japonica , mainly developed in Japan. We combined the target population with three populations that had higher genetic diversity. These were 100 indica accessions, 100 japonica accessions, and 100 accessions with various genetic backgrounds. The results showed that the GWAS's power with a mixed population was generally higher than with a separate population. Also, the optimal GWAS populations varied depending on the fixation index ( F _ST ) of the quantitative trait nucleotides (QTNs) and the polymorphism of QTNs in each population. When a QTN was polymorphic in a target population, a target population combined with a higher diversity population improved the QTN's detection power. By investigating F _ST and the expected heterozygosity ( H _e ) as factors influencing the detection power, we showed that single nucleotide polymorphisms with high F _ST or low H _e are less likely to be detected by GWAS with mixed populations. Sequenced or genotyped germplasm collections can improve the GWAS's detection power by using a subset of the collections with a target population.

Plant response to drought is an important yield-related trait under abiotic stress, but the method for measuring and modeling plant responses in a time series has not been fully established. The objective of this study was to develop a method to measure and model plant response to irrigation changes using time-series multispectral (MS) data. We evaluated 178 soybean (Glycine max (L.) Merr.) accessions under three irrigation treatments at the Arid Land Research Center, Tottori University, Japan in 2019, 2020 and 2021. The irrigation treatments included W5: watering for 5 d followed by no watering 5 d, W10: watering for 10 d followed by no watering 10 d, D10: no watering for 10 d followed by watering 10 d, and D: no watering. To capture the plant responses to irrigation changes, time-series MS data were collected by unmanned aerial vehicle during the irrigation/non-irrigation switch of each irrigation treatment. We built a random regression model (RRM) for each of combination of treatment by year using the time-series MS data. To test the accuracy of the information captured by RRM, we evaluated the coefficient of variation (CV) of fresh shoot weight of all accessions under a total of nine different drought conditions as an indicator of plant’s stability under drought stresses. We built a genomic prediction model (MTRRM model) using the genetic random regression coefficients of RRM as secondary traits and evaluated the accuracy of each model for predicting CV. In 2020 and 2021,the mean prediction accuracies of MTRRM models built in the changing irrigation treatments (r = 0.44 and 0.49, respectively) were higher than that in the continuous drought treatment (r = 0.34 and 0.44, respectively) in the same year. When the CV was predicted using the MTRRM model across 2020 and 2021 in the changing irrigation treatment, the mean prediction accuracy (r = 0.46) was 42% higher than that of the simple genomic prediction model (r =0.32). The results suggest that this RRM method using the time-series MS data can effectively capture the genetic variation of plant response to drought.

In plant breeding programs, rapid production of novel varieties is highly desirable. Genomic selection allows the selection of superior individuals based on genomic estimated breeding values. However, it is worth noting that superior individuals may not always be superior parents. The choice of the crossing pair significantly influences the genotypic value of the resulting progeny. This study has introduced a new crossing strategy, termed cross potential selection (CPS), designed to expedite the production of novel varieties of inbred crops. CPS integrates fast recurrent selection and usefulness criterion to generate novel varieties. It considers the segregation of each crossing pair and computes the expected genotypic values of the top-performing individuals, assuming that the progeny distribution of genotypic values follows a normal distribution. It does not consider genetic diversity and focuses only on producing a novel variety as soon as possible. We simulated a 30-year breeding program in two scenarios, low heritability (h² = 0.3) and high heritability (h² = 0.6), to compare CPS with two other selection strategies. CPS consistently demonstrated the highest genetic gains among the three strategies in early cycles. In the 3^rd year of the breeding program with a high heritability (h² = 0.6), CPS exhibited the highest genetic gains, 138 times that of 300 independent breeding simulations. Regarding long-term improvement, the other selection strategies outperformed CPS. Nevertheless, compared with the other two strategies, CPS achieved significant short-term genetic improvements. CPS is a suitable breeding strategy for the rapid production of varieties within limited time and cost.

Red perilla is an important medicinal plant used in Kampo medicine. The development of elite varieties of this species is urgently required. Medicinal compounds are generally considered target traits in medicinal plant breeding; however, selection based on compound phenotypes (i.e., conventional selection) is expensive and time consuming. Here, we propose genomic selection (GS) and marker-assisted selection (MAS), which use marker information for selection, as suitable selection methods for medicinal plants, and we evaluate the effectiveness of these methods in perilla breeding. Three breeding populations generated from crosses between one red and three green perilla genotypes were used to elucidate the genetic mechanisms underlying the production of major medicinal compounds using quantitative trait locus analysis and evaluating the accuracy of genomic prediction (GP). We found that GP had a sufficiently high accuracy for all traits, confirming that GS is an effective method for perilla breeding. Moreover, the three populations showed varying degrees of segregation, suggesting that using these populations in breeding may simultaneously enhance multiple target traits. This study contributes to research on the genetic mechanisms of the major medicinal compounds of red perilla, as well as the breeding efficiency of this medicinal plant.