European Journal of Sustainable Agroecosystems

Introduction: Abiotic and biotic stressors remain among the main limiting factors affecting global agricultural production, particularly drought, which has been correlated with decreases in average yield of over 10 percent for cereals. The arrival of both high-throughput sequencing and multi-omics technologies has generated an unprecedented amount of information in terms of genomic, transcriptomic, proteomic, and metabolomic datasets, thus providing the opportunity to identify stress-tolerant genes, but also posing a computational challenge when processing and analyzing this data.
Purpose: This paper aims to summarize the existing bioinformatics techniques used for the identification of stress-related genes in crops, which include the methods of differential expression analysis, gene co-expression networks, artificial intelligence and deep learning methods, genome-wide association studies (GWAS), and CRISPR/Cas9-based functional validation, comparing their strengths, weaknesses, and their ability to work together.
Methods: A systematic search on Scopus, Web of Science, PubMed, and Google Scholar was performed, with the time frame of 2015 to 2020 taken into account, along with the grey literature from the Food and Agriculture Organization and international genome-sequencing consortia. From 1,842 references initially found, only 41 studies passed the relevancy criteria and were selected for the detailed analysis via systematic screening.
Key discoveries: The literature review reveals a significant transition in research from studies focused on genes and based on hypotheses to data-driven methods that involve taking advantage of RNA sequencing information together with WGCNA and the use of ensemble approaches such as machine learning for gene selection purposes. The combination of GWAS and multi-environment transcriptomics turns out to be effective in validating multiple candidates. Furthermore, CRISPR/Cas9 technology has established itself as the most common technique in functional validation. Moreover, some transcription factor families (DREB, WRKY, NAC, MYB, and bZIP) can be found in research devoted to drought and salinity stresses on rice, wheat, maize, and soybean cultures, which suggests that certain regulatory mechanisms may have been preserved throughout the evolution of cereals and legumes.
Research limitations: The challenges encountered include the absence of exhaustive field validation of prioritized genes, variability of benchmarking, difficulties in interpreting machine learning results, and underrepresentation of orphan crops.
Concluding remarks: The discovery of genes using bioinformatics techniques has become an important part of climate-resistant crop development nowadays. However, taking full advantage of the prioritization achieved with the help of bioinformatics requires the use of data from various fields and high-standard.