You enter a pollutant. GeneFence branches through a degradation tree and returns a ranked, safety-gated shortlist of ways to clean it up — each one a hypothesis to test, with its reasoning shown.
Every stage runs a real engine, with a graceful fallback when a heavy tool isn't available. The output is ranked hypotheses — never a certified answer.
Your pollutant (a catalog name or a SMILES string) is mapped to a precise structure: canonical SMILES, an InChIKey (the stable fingerprint used for caching), its functional groups, and any priority-pollutant flags.
A rule engine applies known enzymatic transformations (hydrolysis, dehalogenation, ring cleavage, oxidation…) to build a tree of real intermediate structures — the possible ways the molecule could be taken apart.
Each reaction is matched to the enzymes that catalyse it and the organisms known to carry them — and whether that chemistry is tractable in our engineering chassis, Pseudomonas putida.
A genome-scale metabolic model checks whether the host organism stays viable while carrying the new enzymes, and estimates the metabolic burden. Routes that need a specialist anaerobe are flagged infeasible in this chassis.
Each intermediate is checked for toxic substructures (epoxides, vinyl halides, aromatic amines…). Crucially, the model abstains when a molecule is outside what it can reliably judge — rather than guessing.
The escape model estimates whether the engineered construct could spread through the surrounding microbial community via horizontal gene transfer — comparing plasmid vs. chromosomal delivery and naming at-risk neighbours. It also surfaces containment options and the regulatory posture (TSCA TERA/MCAN).
Each root-to-leaf path is one candidate strategy. It gets a composite score from weighted factors — is the host feasible, how efficient the enzyme fit, does it fully mineralize the pollutant — minus penalties for toxic byproducts and gene-escape risk. The breakdown is shown on every result so you can see why it ranked where it did.
Safety is a gate, not just a number. Before ranking, each candidate gets a verdict:
PASS clears the safety checks CAUTION works, but watch a transient toxic intermediate or escape risk VETO makes something worse / can't be contained — sunk to the bottom and hidden by default
The most important thing GeneFence does is refuse to bluff:
GeneFence is decision-support for researchers. Computation proposes and ranks; the wet lab validates, and regulators decide. Nothing here is a recommendation to release an organism.