We describe the development and application of the Sorghum QTL Atlas, a high-resolution, open-access research platform to facilitate candidate gene identification across three cereal species, sorghum, maize and rice. Abstract The mechanisms governing the genetic control of many quantitative traits are only poorly understood and have yet to be fully exploited. Over the last two decades, over a thousand QTL and GWAS studies have been published in the major cereal crops including sorghum, maize and rice. A large body of information has been generated on the genetic basis of quantitative traits, their genomic location, allelic effects and epistatic interactions. However, such QTL information has not been widely applied by cereal improvement programs and genetic researchers worldwide. In part this is due to the heterogeneous nature of QTL studies which leads QTL reliability variation from study to study. Using approaches to adjust the QTL confidence interval, this platform provides access to the most updated sorghum QTL information than any database available, spanning 23 years of research since 1995. The QTL database provides information on the predicted gene models underlying the QTL CI, across all sorghum genome assembly gene sets and maize and rice genome assemblies and also provides information on the diversity of the underlying genes and information on signatures of selection in sorghum. The resulting high-resolution, open-access research platform facilitates candidate gene identification across 3 cereal species, sorghum, maize and rice. Using a number of trait examples, we demonstrate the power and resolution of the resource to facilitate comparative genomics approaches to provide a bridge between genomics and applied breeding.

The yield of cereal crops such as sorghum (Sorghum bicolor L. Moench) depends on the distribution of crop-heads in varying branching arrangements. Therefore, counting the head number per unit area is critical for plant breeders to correlate with the genotypic variation in a specific breeding field. However, measuring such phenotypic traits manually is an extremely labor-intensive process and suffers from low efficiency and human errors. Moreover, the process is almost infeasible for large-scale breeding plantations or experiments. Machine learning-based approaches like deep convolutional neural network (CNN) based object detectors are promising tools for efficient object detection and counting. However, a significant limitation of such deep learning-based approaches is that they typically require a massive amount of hand-labeled images for training, which is still a tedious process. Here, we propose an active learning inspired weakly supervised deep learning framework for sorghum head detection and counting from UAV-based images. We demonstrate that it is possible to significantly reduce human labeling effort without compromising final model performance (R2 between human count and machine count is 0.88) by using a semitrained CNN model (i.e., trained with limited labeled data) to perform synthetic annotation. In addition, we also visualize key features that the network learns. This improves trustworthiness by enabling users to better understand and trust the decisions that the trained deep learning model makes.

Sorghum (Sorghum bicolor L. Moench) is a C4 tropical grass that plays an essential role in providing nutrition to humans and livestock, particularly in marginal rainfall environments. The timing of head development and the number of heads per unit area are key adaptation traits to consider in agronomy and breeding but are time consuming and labor intensive to measure. We propose a two-step machine-based image processing method to detect and count the number of heads from high-resolution images captured by unmanned aerial vehicles (UAVs) in a breeding trial. To demonstrate the performance of the proposed method, 52 images were manually labeled; the precision and recall of head detection were 0.87 and 0.98, respectively, and the coefficient of determination (R2) between the manual and new methods of counting was 0.84. To verify the utility of the method in breeding programs, a geolocation-based plot segmentation method was applied to pre-processed ortho-mosaic images to extract >1000 plots from original RGB images. Forty of these plots were randomly selected and labeled manually; the precision and recall of detection were 0.82 and 0.98, respectively, and the coefficient of determination between manual and algorithm counting was 0.56, with the major source of error being related to the morphology of plants resulting in heads being displayed both within and outside the plot in which the plants were sown, i.e., being allocated to a neighboring plot. Finally, the potential applications in yield estimation from UAV-based imagery from agronomy experiments and scouting of production fields are also discussed.