Comprehensive meta-analysis and machine learning approaches identified the role of novel drought specific genes in Oryza sativa

Abstract

Rice is a major food crop and provides nutrition for half of the world's population. Rice production is majorly affected by drought at different developmental stages and accounted for annual yield loss depending on the intensity of drought. Hence, the need to study the molecular mechanism in a holistic manner behind drought tolerance is a prerequisite to mitigating this problem. Therefore, in the current study, the drought tolerance mechanism of rice plants was elucidated through a meta-analysis on the publically available transcriptomic datasets by integrating these datasets using a R package to remove the batch effects and applying machine learning approaches for prediction robustness and accuracy. Thus, the classifier model identified 128 essential genes through feature selection algorithms and classification methods on training datasets. The comprehensive study revealed that Naïve Bayes classification and correlation-based feature selection was robust in the prediction of essential genes. The accuracy and performance of the classification model was validated with the independent test dataset and the prediction accuracy of the classifier was 93% with ROC (0.972) and F-measures (0.927). Further, the biological significance of the identified genes in drought tolerance was assessed. The current analysis highlighted the regulatory roles of novel genes such as Os01g0844300, Os06g0246500, Os05g03733900, Os05g0550600 Os08g0442900, Os08g0104400, Os01g0256500, Os02g0259900 and Os05g0572700 in the enhancement of drought tolerance mechanisms. Thus the identified genes might be the potential targets for molecular breeding of drought-tolerant rice cultivars.