Machine learning and networks in the analysis of 16S rRNA and ITS sequencing to evaluate the impact of biological additives in mineral fertilizers on soil biodiversity

• Grant number: 24/16862-6
• Support Opportunities: Scholarships abroad - Research Internship - Master's degree
• Start date: January 31, 2025
• End date: July 30, 2025
• Field of knowledge: Agronomical Sciences - Agronomy - Soil Science
• Principal Investigator: Fernando Dini Andreote
• Grantee: Paloma Barros Dias
• Supervisor: Vanessa Nessner Kavamura Noguchi
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil
• Institution abroad: Rothamsted Research, England
• Associated to the scholarship: 23/17001-1 - Effects of biological additive in mineral fertilizer and its relationship with soil biodiversity, BP.MS

Abstract

The pursuit of more sustainable agriculture, driven by increasing pressures on production and consumption standards, demands innovations in soil management and fertilizer use. While mineral fertilizers increase agricultural productivity, their use can compromise soil microbial biodiversity, which is crucial for ecosystem functions and plant development. This study aims to investigate the impacts of combining mineral fertilizers and biological additives on soil microbial biodiversity through the analysis of 16S rRNA and ITS amplicon sequencing data. The data will come from experiments conducted in Brazil, with control treatments and mineral fertilizers, with and without biological additives, applied to soils under different management practices. The analyses will be carried out at Rothamsted Research (UK), using advanced machine learning and network techniques, supported by a high-performance computer (HPC). Algorithms such as SVM and Random Forest will be employed to classify the sequences, comparing them with databases such as SILVA and UNITE. Microbial diversity will be analyzed using PCA and Random Forest, while functional potential will be predicted using Picrust2. Co-occurrence networks and neural networks, together with networks analyses developed from the iNAP pipeline, will allow visualization of ecological interactions, providing a detailed assessment of microbial connectivity and interactions. This study is expected to provide deeper insights into the biological mechanisms associated with the use of fertilizers and biological additives, contributing to the advancement of more sustainable agricultural practices, thus addressing global demand.