Application of AI for the functional elucidation of rice associated microbial community for the improved productivity

Rice microbiome plays a critical role in the growth, health, stress tolerance, nutrient uptake, root development, and productivity of its host. In this study, advanced machine learning algorithms were applied to analyze the genomic data from 1365 rice-associated bacteria sourced from Bacterial and Viral Bioinformatics Resource Center (BV-BRC) database. After filtering, the genomic data of 280 organisms were selected and annotated to identify their respective genes. These were further categorized into ortholog groups, and based on the presence and absence of the ortholog groups, the organisms were clustered into eight groups. Genes encoding amino acid transport, inorganic ion transport and metabolism were the most common Clusters of Orthologous Genes (COG) categories observed across the various clusters while cellular process, biological regulation, and response to stimuli were the most common gene ontology terms. However, the presence of a large proportion of genes having unknown functions suggests the distribution of novel genes which could facilitate the functions including the plant colonization. Further to this, machine learning models were used to classify the organisms as either beneficial or pathogenic. Here, Support Vector Machine based analysis showed the highest accuracy (92.98%) when compared to the Logistic Regression (90.16%) and Random Forest (57.80%). From the analysis, ABC-type transporters such as ABC-type oligopeptide transport system were more abundantly distributed in beneficial bacteria. On the other hand, transposase such as Transposase InsA were observed to be common among pathogenic strains. From the results obtained, the presence of genes responsible for the nutrient transport and metabolic versatility was found to be significant for the beneficial bacteria, while the genetic variability was remarkable for the pathogens. The information generated in this study hence highlights the power of AI for predicting the beneficial interactions between the rice and its microbiome, and thereby offer its applications in enhancing the crop resilience and productivity for the sustainable agricultural practices.

The online version contains supplementary material available at 10.1007/s13205-025-04665-z.