Omnix
Omnix runs predictive machine learning, including antibiotic-resistance prediction grounded in established resistance-gene databases. Every prediction is explainable, and every output is marked research-only.
Inputs
- Analytical results or genomic data from Novora
- Prediction task (including AMR genotype-to-phenotype)
- Reference database selection
Outputs
- Predictions with mandatory SHAP feature attributions
- AMR resistance calls referenced to source databases
- IntendedUse.RESEARCH_ONLY annotation on every output
How the AI suggests, and where you decide
Omnix predicts and explains; it never auto-acts on a prediction. SHAP attributions accompany every output so the researcher can see why a call was made before deciding what to do with it.
Illustrative
Omnix in practice
Citation grounding
AMR predictions reference CARD, ResFinder, AMRFinderPlus, PointFinder, and BV-BRC. SHAP interpretability is mandatory, not optional, on every prediction.
For the technical reader
Technical details
- AMR prediction integrating CARD, ResFinder, AMRFinderPlus, PointFinder, BV-BRC.
- Mandatory SHAP interpretability on every prediction.
Researchers, Clinical research, Public health, Pharma enterprise. The decision-support boundary is surfaced consistently: the platform suggests; the user decides.
Each capability is one link in the closed loop. See the whole chain end to end.