Poncheewin, W. | bioRxiv Science

bioRxiv2021

Classification of the plant-associated phenotype of Pseudomonas strains using genome properties and machine learning

The rhizosphere, the region of soil surrounding roots of plants, is colonized by a unique population of Plant Growth Promoting Rhizobacteria (PGPR). By enhancing nutrient uptake from the soil and through modulation of plant phytohormone status and metabolism, PGPR can increase the stress tolerance, growth and yield of crop plants. Many important PGPR as well as plant pathogens belong to the genus Pseudomonas. There is, however, uncertainty on the divide between phytobeneficial and phytopathogenic strains as previously thought to be signifying genomic features have limited power to separate these strains. Here the Genome properties (GP) common biological pathways annotation system was applied to establish the relationship between the genome wide GP composition and the plant-associated phenotype of 91 Pseudomonas strains representing both phenotypes. GP enrichment analysis, Random Forest model fitting and feature selection revealed 28 discriminating features. A validation dataset of 67 new strains confirmed the importance of the selected features for classification. A number of unexpected discriminating features were found, suggesting involvement of novel molecular mechanisms. The results suggest that GP annotations provide a promising computational tool to better classify the plant-associated phenotype. Author summaryWith a growing population the need to double the agricultural food production is specified. Simultaneously, there is an urgent need to implement sustainable and climate change resilient agricultural practices that preserve natural ecosystems. Cooperative microbiomes play important positive roles in plant growth development and fitness. Properly tuned, these microbiomes can significantly reduce the need for synthetic fertilizers and can replace chemicals in crop pest control. To select beneficial candidates, their traits need to be described and likewise, potential detrimental traits should be avoided. Here we applied GP-based comparative functional genomics, enrichment analysis and Random Forest model fitting to compare known phytobeneficial and phytopathogenic Pseudomonas strains. A number of unexpected discriminating features were found suggesting the involvement of novel molecular mechanisms.

genomics↗

bioRxiv2021

Machine learning approaches to predict the plant-associated phenotype of Xanthomonas strains

The genus Xanthomonas has long been considered to consist predominantly of plant pathogens, but over the last decade there has been an increasing number of reports on non-pathogenic and endophytic members. As Xanthomonas species are prevalent pathogens on a wide variety of important crops around the world, there is a need to distinguish between these plant-associated phenotypes. To date a large number of Xanthomonas genomes have been sequenced, which enables the application of machine learning (ML) approaches on the genome content to predict this phenotype. Until now such approaches to the pathogenomics of Xanthomonas strains have been hampered by the fragmentation of information regarding strain pathogenicity over many studies. Unification of this information into a single resource was therefore considered to be an essential step. Mining of 39 papers considering both plant-associated phenotypes, allowed for a phenotypic classification of 578 Xanthomonas strains. For 65 plant-pathogenic and 53 non-pathogenic strains the corresponding genomes were available and de novo annotated for the presence of Pfam protein domains used as features to train and compare three ML classification algorithms; CART, Lasso and Random Forest. Recursive feature extraction provided further insights into the virulence enabling factors, but also yielded domains linked to traits not present in pathogenic strains.

microbiology↗