A new virus emerges. We start with its cavities.
When a novel pathogen or variant appears, there is a genome long before there are structures, targets or drugs. Innophore’s Catalophore™ point-cloud technology maps every binding site across a viral proteome — to repurpose approved drugs, design broad-spectrum antivirals and track variant escape, in days rather than years.
Outbreaks move in weeks. Antiviral pipelines move in years.
A new virus or variant is sequenced within days — but a sequence is not a drug target. Developing a novel antiviral takes 10–15 years, and roughly 30% of trials fail over side effects. The bottleneck is the binding site: which viral pockets are druggable, which approved drugs already fit them, and how those pockets shift as the virus mutates.
Sequence, no structure
An emerging pathogen yields thousands of open reading frames, but no experimental structures and no validated targets — the starting point for any therapy is missing.
Druggable cavities
Viral targets often share little sequence or fold with anything characterized. Sequence search misses the approved drugs whose binding sites already match — and the cross-species pockets that enable broad-spectrum design.
Variants & resistance
Spike mutations reshape the receptor interface; protease mutations erode antiviral efficacy. A target that worked last season can escape the next.
Two proteins that share no evolutionary ancestry can still develop pockets shaped alike. Sequence-based methods never see it — cavity similarity does.
The convergent-cavity principleFrom a viral genome to a searchable map of every binding site.
Catalophore™ predicts and curates a virus’s structural proteome, then describes each binding pocket as a 3D point-cloud signature of its physico-chemical property fields. Those signatures are searched against approved drugs and the proteome — independent of sequence, fold or family — to surface inhibitors, repurposing candidates and off-target risks. The same engine behind our pharma and enzyme work, pointed at pathogens.
For an emerging pathogen, the cavity map is built on AI-predicted structures — and AI structure prediction still gets the chemistry wrong often enough to matter: by one benchmark, chirality errors occur in 4.4% of cases. At proteome scale, that is thousands of subtly wrong binding sites feeding into a drug-discovery decision.
That is why every structure entering a cavitome is energy-minimized, refined and curated — not taken on faith. When the question is which antiviral to advance, the inputs have to be right.
From an emerging genome to an antiviral candidate.
One continuous path — the same response whether it is a new variant or a never-seen pathogen.
Emerging pathogen
A new genome or variant — thousands of open reading frames, no experimental structures, no validated targets.
Cavity analysis
Models and curates the structural proteome, maps every binding site as a point cloud, and searches approved drugs and the proteome for matches — ligand-agnostic.
Antiviral candidates
Repurposed drugs, broad-spectrum leads, variant-aware decoys and prioritized vaccine / drug-target sites.
From the first genome to a drug target — in hours.
When SARS-CoV-2 emerged, we did not wait for a pipeline. We turned cavity science on the live outbreak — in real time, and in public.
In January 2020 — within hours of the first SARS-CoV-2 genome being published — we built a homology model of the viral main protease (Mpro) and screened for inhibitors of its active site. That is the exact target Pfizer’s Paxlovid (nirmatrelvir) would be approved against two years later.
Live structural-bioinformatics outbreak log · 23 January 2020
A public log from the outbreak
The day the genome (GenBank MN908947) was released, we published a time-stamped structural-bioinformatics log — protease models, cavity analyses and candidate inhibitors, openly shared with the research community.
Models the world used
Our open Mpro models were downloaded tens of thousands of times — kickstarting a partnership with the Chinese Center for Disease Control and Prevention together with Pharmaron (Beijing HQ), and used by Insilico Medicine.
A billion molecules against COVID-19
Finalist in the JEDI “Billion Molecules against COVID-19” Grand Challenge, alongside Harvard Medical School (VirtualFlow) and Google Cloud — ultra-large screens across 17 viral targets.
Real-time variant tracking
We screened over 14 million variant proteins on AWS and ran virus.watch, a real-time SARS-CoV-2 variant-monitoring system built with the AWS Diagnostic Development Initiative.
In the press: “Open for outbreaks”, Nature Biotechnology ↗ · “The pandemic pipeline”, Nature Biotechnology ↗ · WIRED, August 2020 ↗
Proven on real pathogens, in peer-reviewed work.
Cavity-based virology is not a concept. It has been applied to SARS-CoV-2, monkeypox and arboviruses — and published.
Approved drugs as antivirals
DrugSolver CavitomiX flagged flufenamic acid and fusidic acid as SARS-CoV-2 inhibitors — despite under 5% sequence and near-zero structural identity to the viral targets — then confirmed them in vitro.
Scientific Reports 13, 11783 (2023) ↗
Antivirals by design
A genetic algorithm refines inhibitors of the chikungunya nsP3 ADP-ribose site for reduced off-target activity and breadth across viral variants and species — safe, broad antivirals for arboviral diseases.
Viruses 16, 1186 (2024) ↗
A whole virus, modeled
A consensus structural proteome of the circulating monkeypox virus: 3,713 genomes distilled into 210 atomistic protein models, with a mutational analysis of the tecovirimat and brincidofovir binding sites.
Microbiology Spectrum 11 (2023) ↗
Scoring spike escape
Structural-bioinformatics scoring of spike–hACE2 affinity across variants of concern flagged Omicron’s elevated receptor binding — building on the earlier discovery that residue S477 governs the interface.
Sci. Rep. 12 (2022) ↗ · Sci. Rep. 11 (2021) ↗
Watching the protease
Tracking the mutational dynamics of the SARS-CoV-2 main protease (Mpro) revealed an emerging danger of resistance to Mpro-targeting antiviral drugs — early warning from structure.
Frontiers in Medicine 9 (2022) ↗
Variant-proof biologics
Deep-learning-guided molecular dynamics optimized soluble hACE2-Fc decoys (e.g. K31W) that capture SARS-CoV-2 regardless of spike mutation — validated by virus-neutralization assays.
Scientific Reports 13 (2023) ↗
A decade of structural virology, run at proteome scale.
The same patented, GPU-accelerated Catalophore™ engine behind our pharma and enzyme work has a deep, published track record on viral targets — from the first weeks of COVID-19 to the 2022 monkeypox outbreak and arboviral threats.
Fusidic acid and flufenamic acid share under 5% sequence identity and almost no structural similarity with the SARS-CoV-2 targets they inhibit. Only a cavity-based search could have found them.
Repurposing beyond sequence & foldCavitOmiX Drug Discovery: engineering antivirals with enhanced spectrum and reduced side effects for arboviral diseases
Viruses 16, 1186 — broad-spectrum, low-off-target antiviral design
A community effort in SARS-CoV-2 drug discovery
Molecular Informatics 43 — international consortium, incl. Innophore
AI-assisted structural consensus-proteome prediction of human monkeypox viruses
Microbiology Spectrum 11 — 210-protein structural proteome of the 2022 outbreak
Identification and validation of fusidic acid and flufenamic acid as inhibitors of SARS-CoV-2 replication using DrugSolver CavitomiX
Scientific Reports 13, 11783 — cavity-based drug repurposing, in-vitro validated
Optimizing variant-specific therapeutic SARS-CoV-2 decoys using deep-learning-guided molecular dynamics simulations
Scientific Reports 13 — variant-proof hACE2-Fc decoys
Recent changes in the mutational dynamics of the SARS-CoV-2 main protease substantiate the danger of emerging resistance to antiviral drugs
Frontiers in Medicine 9 — antiviral resistance early warning
Structural bioinformatics analysis of SARS-CoV-2 variants reveals higher hACE2 receptor binding affinity for Omicron B.1.1.529 spike RBD
Scientific Reports 12 — variant affinity scoring
Serine 477 plays a crucial role in the interaction of the SARS-CoV-2 spike protein with the human receptor ACE2
Scientific Reports 11 — spike–ACE2 interface mechanism
A multi-pronged approach targeting SARS-CoV-2 proteins using ultra-large virtual screening
iScience 24, 102021 — 17 targets, 45 screens, 50 billion docking instances
Selected peer-reviewed publications with Innophore authorship. Full list at Technology › Publications.
An early response, on demand.
CavitOmiX is a pipeline, not a one-off study. Submit a viral target or a whole proteome and retrieve ranked repurposing candidates and inhibitors — through a standard API, ready for your screening and modeling stack.
- API
Standard REST API
Submit viral structures, retrieve ranked cavity matches and repurposing hits.
- I/O
Model-ready outputs
Candidate lists formatted to drop straight into downstream docking and safety pipelines.
- ∞
Pathogen-agnostic
The same workflow for SARS-CoV-2, poxviruses or the next emerging threat.
# scan a viral target for repurposing candidates POST /v1/cavity/scan { "structure": "<viral_target.pdb>", "site": "active-site", "library": "approved-drugs" } # → ranked candidates + rationale { "matches": [ { "drug": "fusidic acid", "similarity": 0.89, "evidence": "cavity" } ] }
Bring us a virus.
Whether you respond to outbreaks or build the platforms behind antiviral discovery — there’s a way in.
Request a pilot
Bring a target or an emerging pathogen. We map its proteome’s cavities and return repurposing and lead candidates with structural evidence.
Start a pilot →Explore integration
Add proteome-wide cavity search to your antiviral or vaccine pipeline via a standard API.
Talk integration →