Tensor Omics is a novel bioinformatics framework that analyzes transcriptome and integrated omics data by projecting it into semantically structured vector spaces, where each axis reflects a biological context such as tissue, diet, disease, or aging. It enables for example geometric exploration of gene families through expression shifts from conserved states, uncovering divergence, specificity, and adaptation. Functional divergence—such as neo- and sub-functionalization or dosage effects—is detected using simple vector algebra, providing a novel method for reliable outlier detection across species and omics layers.
Tensor Omics can be applied to metagenomic transcriptome analysis by quantifying gene expression shifts between related genes and species, and also supports trajectory analysis in time-series data by modeling omics dynamics as curves in high-dimensional space. Unlike traditional systems biology tools, it is inherently extensible to proteomics, metabolomics, and phenomics, enabling intuitive, multi-scale investigations into stress responses, aging, and microbiome function. By operating on vectors, matrices, and tensors—mathematical objects native to AI—it bridges high-dimensional biology with machine learning for scalable, interpretable analysis.
| Talk: | Tensor Omics - Multi Omics integration, semantic geometric analysis, and explainable Biomarker learning | |
| Speaker: | Asis Hallab University of Applied Sciences Bingen and Forschungszentrum Jülich | |
| Date: | Monday, November 3, 2025 at 3 pm | |
| Location: | Seminar room „Nucleus“, FLI 1, Beutenbergstraße 11, Jena | |
| Host: | Dario Riccardo Valenzano |
The seminar will be held as an in-person event.
