AUTHOR=Duhan Naveen , Kaundal Rakesh 

TITLE=PRGminer: harnessing deep learning for the prediction of resistance genes involved in plant defense mechanisms

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2025.1411525

DOI=10.3389/fpls.2025.1411525

ISSN=1664-462X

ABSTRACT=Plant resistance genes are crucial in plant defense systems against a variety of diseases and pests. These plant-specific genes encode proteins that identify particular molecular patterns associated with pathogens invading the plants. When these resistance genes are active, they initiate a sequence of molecular processes that culminate in the activation of defensive responses such as the synthesis of antimicrobial chemicals, cell wall strengthening, and triggering of programmed cell death in infected cells. Plant resistance genes are exceedingly varied, with several classes and subclasses found across a wide range of plant species. The identification of new resistance genes (Rgenes) is a critical component of disease resistance breeding. Nonetheless, identifying Rgenes in wild species and near relatives of plants is not only challenging but also time-consuming. In this study, we present PRGminer, a deep learning-based high-throughput Rgenes prediction tool. PRGminer is implemented in two phases: Phase I predicts the input protein sequences as Rgenes or non-Rgenes; and Phase II classify the Rgenes predicted in Phase I into eight different classes. Among all the sequence representations tested, the dipeptide composition gave the best prediction performance (accuracy of 98.75% in a k-fold training/testing procedure, and 95.72% on an independent testing) with a high Matthews correlation coefficient (0.98 training and 0.91 in independent testing) in Phase I; phase II (overall accuracy of 97.55% in a k-fold training/testing and 97.21% in an independent testing) with the MCC values of 0.93 for k-fold training procedure and 0.92 in an independent testing. PRGminer is available as a webserver which can be freely accessed at https://kaabil.net/prgminer/, as well as a standalone tool available for download at https://github.com/usubioinfo/PRGminer. PRGminer will help researchers to accelerate the discovery of new R genes, understand the genetic basis of plant resistance, and develop new strategies for breeding plants that are resistant to disease and pests.