Looking to join consortium for HORIZON-MISS-2026-05-SOIL-04

Bar-Ilan University (BIU) is one of Israel’s leading research universities with a strong track record in interdisciplinary research. The EnICS Labs (Emerging Nanoscaled Circuits and Systems) at the Faculty of Engineering specializes in AI hardware acceleration, processing-in-memory (PiM), and content-addressable memory (CAM) architectures. The group has extensive, demonstrated expertise in AI-based metagenomic analysis of soil microbiomes, including the development of deep learning pipelines (transformer, CNN, RNN, and state space models) for taxonomic classification and functional annotation of complex microbial communities in environmental samples. EnICS Labs has developed proprietary computing platforms and taped out over 40 chips, including dedicated AI processors for bioinformatics.

We bring directly relevant, ongoing research in AI-based metagenomic analysis of soil microbiomes, which we propose to integrate into the project’s AI-driven decision support system for sustainable soil management. Our specific contributions include:

(1) AI-powered soil microbiome analysis as a soil health indicator: We are developing a metagenomic analysis platform combining transformer-based models, CNNs, RNNs, and state space models for taxonomic classification and functional annotation of complex soil microbial communities. This platform can process shotgun and long-read sequencing data (Oxford Nanopore, PacBio) from soil samples collected at Long-Term Field Experiment (LTE) sites, providing microbial diversity and functional capacity indicators directly relevant to soil health assessment and monitoring.

(2) Open-source AI/ML tools for soil data analytics: We will develop open-source, modular AI components for analyzing complex, heterogeneous soil datasets, including metagenomic, physicochemical, and agronomic data from LTE networks. Our self-supervised learning approach enables taxonomy-free classification of novel microbial species (the “dark matter” of soil metagenomes), overcoming the limitation of incomplete reference databases that currently hinder soil microbiome characterization.

(3) Real-time, field-deployable analysis for soil management decisions: Our platform integrates sequencing with deep learning-based real-time analysis, enabling adaptive soil monitoring. This capability can predict pollutant degradation pathways, antimicrobial resistance (AMR) markers, microbial community shifts, and biodegradation potential, providing actionable indicators for land managers and advisors making soil management decisions.

(4) Predictive modelling for soil health: Using our expertise in de novo metagenomic assembly, probabilistic binning, and deep learning-based genome classification, we will contribute to building predictive models that link microbial community composition and functional profiles to soil health indicators, supporting evidence-based sustainable soil management strategies across pedo-climatic regions.

We are currently developing an AI-based metagenomic analysis platform specifically designed for contaminated soil microbiomes, which can be directly extended and adapted for the sustainable soil management objectives of this call.