Spatial and temporal microbiome profiling for precision farming and land segmentation
A research program combining omics-based spatial profiling of plant root microbiomes with a novel spatially-aware machine learning algorithm (KODAMA) to enable micro-scale agricultural interventions. Targets precision application of fertilizers and pesticides in crops like Sorghum, reducing environmental impact and costs across African agroecosystems.
Overview
This program addresses spatial heterogeneity in soil and plant properties—a key driver of biogeochemical cycling, nutrient losses, and ecosystem function. By combining microbiome profiling of Sorghum roots with spatially-aware machine learning, the research enables targeted, micro-scale agricultural interventions such as precise fertilizer or pesticide application. The approach supports precision farming by characterizing how management practices influence spatial patterns in soil and plant properties, ultimately allowing more targeted crop management recommendations that reduce environmental impact and costs.
Technical specifications
Core capabilities:
- Microbiome profiling: Long-read sequencing using Nanopore technology to profile microbial communities in Sorghum roots, with over 1,000 bulk soil and rhizosphere samples collected before and after pesticide treatment across multiple African regions
- Metabolomics and elemental analysis: NMR-based untargeted metabolomics for macro-nutrient profiling and ICP-MS for elemental analysis of soil samples
- Spatial data integration: A modified version of KODAMA, an unsupervised machine-learning algorithm for feature extraction from noisy, high-dimensional data, enhanced to incorporate spatial coordinates so that predictions consider spatially neighboring samples
- Land segmentation: Novel spatially-aware unsupervised method for segmenting agricultural land based on integrated omic and spatial data, validated against benchmark spatial transcriptomics datasets including Visium technology
- Visualization: Low-dimensional embedding of results using t-SNE or UMAP for intuitive interpretation
The KODAMA algorithm is currently available as an R package on CRAN, with development versions maintained on public repositories.
Technology readiness level
The program is currently at an active research and development stage. Sample collection is underway across commercial and smallholder farms in South Africa, Kenya, and Uganda. The KODAMA algorithm has been validated in prior work and is publicly available, while the spatially-aware extension is under development and will be validated using established spatial transcriptomics benchmarks. The research is supported by a collaborative network across ICGEB's Bioinformatics, Plant Systems Biology, and Biopesticides groups, with potential for pilot deployment in precision agriculture contexts following algorithm validation.
About International Centre for Genetic Engineering and Biotechnology
The International Centre for Genetic Engineering and Biotechnology is an intergovernmental, non-profit research and training organization operating laboratories in Trieste, New Delhi, and Cape Town. Its campuses provide co-located labs and shared core facilities that enable collaborative R&D, while training and fellowship programs create sustained talent pipelines for partners. Proximity to innovation parks and leading universities at each site, plus biosafety and policy advisory services, helps industry navigate translation and regulatory contexts across regions. Research is supported by member-state contributions and competitive grants from national and international agencies. A dedicated technology transfer office manages IP, licensing, and commercialization across sites.
delivered to your inbox.