Computational biology for decarbonization
Computational biology for decarbonization.
Stoma turns unannotated genomes into ranked, testable hypotheses — telling a wet-lab biologist exactly where to look first. The same foundation is built to grow into engineering: turning what the platform learns into decarbonization technology, not just better search.
Novel methods in our pipeline
- AlphaFold
- ESM-2
- Foldseek
- InterPro
- SignalP
Approach
Decarbonization is a biology problem, not an energy problem.
The actual constraint is biological knowledge, not engineering throughput. A long list of decarbonization-relevant industrial processes is bottlenecked by organisms that already do this work in nature — non-model eukaryotes, symbiotic systems, lineages with no close relative in any reference database. Most of their genes have no known function.
Most computational tools for this kind of work inherit assumptions from drug-discovery pipelines: well-characterized model organisms, dense reference data, large internal datasets. Non-model biology has none of that — closing the gap means building for data scarcity directly, not adapting tools built for a different problem.
Stoma is structured around two linked steps: a platform that makes a finite amount of wet-lab time go further, and an engineering commitment that the platform's findings are meant to eventually inform. The platform funds and informs the engineering — on purpose, from day one — rather than the two running as separate bets that happen to share a name. Every ranked hypothesis the platform produces is a candidate input for what engineering eventually builds.
Platform
A ranked list of what's worth testing next.
One pipeline, one ranked output — instead of five disconnected tools a biologist has to run and reconcile by hand.
Unannotated genome
Thousands of genes, no assigned function
Five-method pipeline
- Function class
- Structural homology
- Interaction partners
- Transit peptides
- Regulatory signals
Ranked hypothesis list
Confidence-scored, not verified
Wet lab
Where the proving happens
What the output actually is
Every output is a ranked hypothesis with a confidence score — not a verified function. This is a method and architecture demonstration, not a calibrated accuracy claim. It tells you where to look first. The wet lab still does the proving.
Get in touch