30B Parameters
Large-scale biological foundation model
From DNA sequence to commercial-scale manufacturing. Canopy is twig bio’s graph-native foundation AI model for engineering biology, designed to predict how microbes will perform before they are ever built.
By reasoning across biological sequences, metabolic pathways, host organisms, fermentation conditions and proprietary experimental data, Canopy enables faster, more reliable development of industrial bioprocesses.
Large-scale biological foundation model
Trained across diverse organisms
Per programme on average
across industrial programmes
Spanning distinct product classes
Cross scale reasoning and biological design From biological complexity to manufacturing outcomes
Amino acid sequences and structures
Pathways and molecular interactions
Microbial diversity from environments
Genomics, transcriptomics, proteomics, metabolomics
High-throughput assays and process data
Discover and design enzymes with novel functions
Design and optimize pathways to target molecules
Engineer robust microbial cell factories
Protect yield, titer and productivity across conditions
Generate actionable insights for scale-up/process control
Traditional Design-Build-Test-Learn workflows are slow, expensive and highly iterative. Scientists spend months screening thousands of variants before discovering bottlenecks, resulting in excessive cost, long timelines and significant technical risk.
Canopy supports the complete strain engineering workflow and feeds every prediction back into a closed Design-Build-Test-Learn cycle.
Design enzymes for reactions where no suitable catalyst exists.
Generate biosynthetic pathways from central metabolism to target molecule.
Rank knockouts, overexpression strategies and CRISPR edits for maximum production.
Predict titre, productivity and process behaviour across manufacturing conditions.
Independent benchmarking confirms strong predictive capability, underpinned by a foundation model trained at unprecedented biological scale.
Canopy combines four core capabilities into a single, unified platform that reasons across the full biological stack.
Independent benchmarking confirms strong predictive capability, underpinned by a foundation model trained at unprecedented biological scale.
Titre Prediction — accurate forecasting of continuous fermentation performance across held-out datasets.
Expression Classification — reliable identification of high-performing production strains before lab work.
Successfully prioritises genetic interventions before laboratory validation, reducing experimental cost.
Differentiates construct architectures using whole-plasmid sequence representations.
Independent benchmarking confirms strong predictive capability, underpinned by a foundation model trained at unprecedented biological scale
CANOPY is being developed with support from the UK Sovereign AI programme, helping establish strategic national capability in AI for engineering biology and biomanufacturing.