Plant diseases pose a significant challenge for food production and safety. Therefore, it is indispensable to correctly identify plant diseases for timely intervention to protect crops from massive losses. The application of computer vision technology in phytopathology has increased exponentially due to automatic and accurate disease detection capability. However, a deep convolutional neural network (CNN) requires high computational resources, limiting its portability. In this study, a lightweight convolutional neural network was designed by incorporating different attention modules to improve the performance of the models. The models were trained, validated, and tested using tomato leaf disease datasets split into an 8:1:1 ratio. The efficacy of the various attention modules in plant disease classification was compared in terms of the performance and computational complexity of the models. The performance of the models was evaluated using the standard classification accuracy metrics (precision, recall, and F1 score). The results showed that CNN with attention mechanism improved the interclass precision and recall, thus increasing the overall accuracy (>1.1%). Moreover, the lightweight model significantly reduced network parameters (~16 times) and complexity (~23 times) compared to the standard ResNet50 model. However, amongst the proposed lightweight models, the model with attention mechanism nominally increased the network complexity and parameters compared to the model without attention modules, thereby producing better detection accuracy. Although all the attention modules enhanced the performance of CNN, the convolutional block attention module (CBAM) was the best (average accuracy 99.69%), followed by the self-attention (SA) mechanism (average accuracy 99.34%).

Background Cold stress is the main abiotic stress in rice, which seriously affects the growth and yield of rice. Identification of cold tolerance genes is of great significance for rice to solve these problems. GATA-family transcription factors involve diverse biological functions, however, their role in cold tolerance in rice remains unclear. Results In this study, a GATA-type zinc finger transcription factor OsGATA16, which can improve cold tolerance, was isolated and characterized from rice. OsGATA16 belongs to OsGATA subfamily-II and contains 11 putative phosphorylation sites, a nuclear localization signal (NLS), and other several conserved domains. OsGATA16 was expressed in all plant tissues, with the strongest in panicles. It was induced by cold and ABA treatments, but was repressed by drought, cytokinin and JA, and acted as a transcriptional suppressor in the nucleus. Overexpression of OsGATA16 improves cold tolerance of rice at seedling stage. Under cold stress treatments, the transcription of four cold-related genes OsWRKY45–1, OsSRFP1, OsCYL4, and OsMYB30 was repressed in OsGATA16-overexpressing (OE) rice compared with wild-type (WT). Interestingly, OsGATA16 bound to the promoter of OsWRKY45–1 and repressed its expression. In addition, haplotype analysis showed that OsGATA16 polarized between the two major rice subspecies japonica and indica, and had a non-synonymous SNP8 (336G) associated with cold tolerance. Conclusion OsGATA16 is a GATA transcription factor, which improves cold tolerance at seedling stage in rice. It acts as a positive regulator of cold tolerance by repressing some cold-related genes such as OsWRKY45–1, OsSRFP1, OsCYL4 and OsMYB30. Additionally, OsGATA16 has a non-synonymous SNP8 (336G) associated with cold tolerance on CDS region. This study provides a theoretical basis for elucidating the mechanism of cold tolerance in rice and new germplasm resources for rice breeding.

Root network structure plays a crucial role in growth and development processes in rice. Longer, more branched root structures help plants to assimilate water and nutrition from soil, support robust plant growth, and improve resilience to stresses such as disease. Understanding the molecular basis of root development through screening of root-related traits in rice germplasms is critical to future rice breeding programs. This study used a small germplasm collection of 137 rice varieties chosen from the Korean rice core set (KRICE_CORE) to identify loci linked to root development. Two million high-quality single nucleotide polymorphisms (SNPs) were used as the genotype, with maximum root length (MRL) and total root weight (TRW) in seedlings used as the phenotype. Genome-wide association study (GWAS) combined with Principal Components Analysis (PCA) and Kinship matrix analysis identified four quantitative trait loci (QTLs) on chromosomes 3, 6, and 8. Two QTLs were linked to MRL and two were related to TRW. Analysis of Linkage Disequilibrium (LD) decay identified a 230 kb exploratory range for detection of candidate root-related genes. Candidates were filtered using RNA-seq data, gene annotations, and quantitative real-time PCR (qRT-PCR), and five previously characterized genes related to root development were identified, as well as four novel candidate genes. Promoter analysis of candidate genes showed that LOC_Os03g08880 and LOC_Os06g13060 contained SNPs with the potential to impact gene expression in root-related promoter motifs. Haplotype analysis of candidate genes revealed diverse haplotypes that were significantly associated with phenotypic variation. Taken together, these results indicate that LOC_Os03g08880 and LOC_Os06g13060 are strong candidate genes for root development functions. The significant haplotypes identified in this study will be beneficial in future breeding programs for root improvement.

A feasible strategy of on-demand drug delivery for the treatment of dermal inflammation under low-pH conditions is proposed, employing zeolitic imidazolate framework-8 (ZIF-8) as a pH-responsive nanoparticle and curcumin (CCM) as a model drug. To overcome the low bioavailability of topically treated drug, a microneedle (MN) form is used to incorporate CCM and ZIF-8. Taking advantage of the fact that ZIF-8 degrades under acidic conditions, CCM is embedded in porous ZIF-8 nanoparticles such that CCM is released when ZIF-8 comes into contact with an acidic dermal fluid at the inflammation site, and this CCM-encapsulated ZIF-8 (CCMZIF) is incorporated into water-dissolvable poly­(vinyl pyrrolidone) MN. The ZIF-8 shows a high loading capacity (∼40.5%) of CCM through chemical bonding and physical adsorption. From in vitro tests with both a buffered solution and porcine skin, CCM from the CCMZIF MN is released in a higher amount at pH 5.0 than at pH 7.4, demonstrating the capability of the pH-responsive release of the drug when needed at inflammatory sites. The analytical investigation conducted here reveals that an acidic environment triggers the structural degradation of ZIF-8, allowing the release of the chemically bonded CCM. Cytotoxicity and stability tests demonstrate the good biocompatibility and bioavailability of ZIF-8. This study highlights the analytical discussion of the encapsulation and release mechanism of CCM in a ZIF-8-implemented MN drug delivery platform. The results demonstrate an advanced on-demand therapeutic strategy for skin disorder treatment.

Salinity is one of the major constraints causing soil problems and is considered a limitation to increased rice production in rice-growing countries. This genome-wide association study (GWAS) experiment was conducted to understand the genetic basis of salt tolerance at the seedling stage in Korean rice. After 10 days of salt stress treatment, salt tolerance was evaluated with a standard evaluation system using a visual salt injury score. With 191 Korean landrace accessions and their genotypes, including 266,040 single-nucleotide polymorphisms (SNPs), using a KNU Axiom Oryza 580K Genotyping Array, GWAS was conducted to detect three QTLs with significant SNPs with a −log10(P) threshold of ≥3.66. The QTL of qSIS2, showed −log10(P) = 3.80 and the lead SNP explained 7.87% of total phenotypic variation. The QTL of qSIS4, showed −log10(P) = 4.05 and the lead SNP explained 10.53% of total phenotypic variation. The QTL of qSIS8 showed −log10(P) = 3.78 and the lead SNP explained 7.83% of total phenotypic variation. Among the annotated genes located in these three QTL regions, five genes were selected as candidates (Os04g0481600, Os04g0485300, Os04g0493000, Os04g0493300, and Os08g0390200) for salt tolerance in rice seedlings based on the gene expression database and their previously known functions.

No abstract

Grain size affects the yield and quality of rice. The large grain line (LGL), showing a large grain size and japonica-like genome, was selected in the breeding field. The 94 F2 plants derived from a cross between LGL and Hanareum (a high-yielding tongil-type variety) were used for the quantitative trait loci (QTL) analysis of grain length (GL), grain width (GW), and grain thickness (GT). A linkage map of the F2 population, covering 1312 cM for all 12 chromosomes, was constructed using 123 Fluidigm SNP markers. A total of nine QTLs for the three traits were detected on chromosomes two, three, four, six, and seven. Two QTLs for GL on chromosomes two and six explained 17.3% and 16.2% of the phenotypic variation, respectively. Two QTLs were identified for GW on chromosomes two and three, and explained 24.3% and 23.5% of the phenotypic variation, respectively. The five QTLs for GT detected on chromosomes two, three, five, six and seven, explained 13.2%, 14.5%, 16.6%, 10.9%, and 10.2% of the phenotypic variation, respectively. A novel QTL for GT, qGT2, was validated on the same region of chromosome two in the selected F3 population. The QTLs identified in this study, and LGL, could be applied to the development of large-grain rice varieties.

This study aimed to develop a service model that uses a deep learning algorithm for detecting diseases and pests in strawberries through image data. In addition, the pest detection performance of deep learning models was further improved by proposing segmented image data sets specialized in disease and pest symptoms. The CNN-based YOLO deep learning model was selected to enhance the existing R-CNN-based model's slow learning speed and inference speed. A general image data set and a proposed segmented image dataset was prepared to train the pest and disease detection model. When the deep learning model was trained with the general training data set, the pest detection rate was 81.35%, and the pest detection reliability was 73.35%. On the other hand, when the deep learning model was trained with the segmented image dataset, the pest detection rate increased to 91.93%, and detection reliability was increased to 83.41%. This study concludes with the possibility of improving the performance of the deep learning model by using a segmented image dataset instead of a general image dataset.