OmniRayn Discovery · Research-gate

Computational candidate discovery.
From natural products to novel composition.

A curated exchange of mechanism-targeted candidate intelligence, built on a research program that turns genomic and peptide-sequence analysis into candidate assets — one robust polyketide lead, a peptide-candidate portfolio in development. Every candidate is computational; results are reported against null controls, with no compound yet isolated, characterised, or tested.

Who it’s for

Find your path.

Everything below is computational and pending wet-lab — no disease or efficacy claims. Pick the door that fits; the deeper detail (assets, methodology) is available under NDA.

For pharma & BD

Licensable assets

Validation-tiered candidate intelligence; novel-composition leads licensable under NDA; 505(b)(2)/orphan-aware framing.

Licensing & the Exchange →
For academics

Free teaching packages

Free, honest-science teaching packages (L0–L3) — real case studies, clearly labelled “not an efficacy grade.” Academic research & education use is free; escalate to a commercial license → when the work goes commercial (spin-out / product / industry partner) — academic-spin-out rates available.

Open academic access →
For researchers & builders

Whitespace intelligence

Tox-filtered, trial-whitespace-mapped candidate intelligence — computational, pending wet-lab, no efficacy claims.

Browse the Exchange →
7
Genomes Surveyed · Input
131
BGC Candidates · Computational
4
Peptide Candidates · Computational
1
Polyketide Lead · Survived Every Null

Cross-kingdom validation across seven medicinal Basidiomycete and Ascomycete genomes. Every candidate tested against compositional and intergenic null distributions before reporting. Architectural-class and computational evidence only — no pharmacological claims; no compound has been isolated, characterised, or tested. The peptide-candidate portfolio is at the computational-candidate stage with provisional intellectual-property filings.

Pipeline

From genome and sequence to partner-ready candidate.

OmniRayn Discovery advances every asset along one path — computational identification, multi-null validation, structural prediction, then partner-funded wet-lab confirmation. Each asset is shown at the stage it has actually reached. Every asset is produced in-house by a proprietary discovery capability OmniRayn retains; we license assets, not infrastructure.

Asset
Class
Stage
Status
OmniRayn Discovery
Foundation-model biosynthetic-gene-cluster identification
Operational
Live
Peptide portfolio
Proglucagon-family peptide candidates
Research; staged for wet-lab
Research
FirstRayn
Genome-guided functional supplement
DSHEA pathway
Reference

Architectural-class and computational evidence only. No compound has been isolated, characterised, or tested. Asset-level diligence detail is available to qualified partners under NDA.

Null-Control Method

Foundation-model discovery — stress-tested under orthogonal null controls.

Most BGC ranking pipelines distinguish only between “real DNA” and shuffled DNA. We add a second layer that distinguishes BGCs from random non-BGC regions of the same organism's own genome. Most candidates fail the second layer. The ones that survive are the ones we report.

Two-channel null-control framework

A compositional null (dinucleotide-preserving shuffle) tests whether a candidate looks like real genomic sequence at all. An intergenic null (in-distribution windows from the same organism) tests whether the embedding distance actually separates BGCs from random non-BGC genome of the same organism. Most pipelines skip the second; we run both, every candidate, every time.

Foundation-model bridge

Each candidate locus passes through a 7-billion-parameter nucleotide foundation model and is embedded into a high-dimensional pattern space. We do not truncate or PCA-reduce. Why: biosynthetic signal is carried in part by the tail of the hidden state — practitioner truncation to 256 or 512 dims loses information that materially changes the candidate ranking.

Headline result

One robust Type I polyketide candidate in shiitake survived every null control we applied.

Architectural features align with the statin family. We make no claim about pharmacological behavior or compound chemistry — no compound has been isolated, characterised, or tested. The result is a genomic architecture finding, not a drug claim. Full per-cluster detail is in the manuscript currently in submission.

Reviewer copy with deeper detail available under NDA — research@omnirayn.com

How discovery works

We classify. We don’t filter.

Most genome-mining tools keep only candidates that match a known chemical class and quietly discard the rest. We retain and label every candidate the analysis surfaces — including novel-scaffold candidates with no known analog, often the richest source of genuinely new chemistry — then turn the strongest into synthesis- and assay-ready design packages, each labelled with exactly how much is grounded versus hypothesis.

01

Classify, don’t filter

Every candidate is retained and labelled, not discarded for failing to match a known class. The novel-scaffold set — candidates with no known analog — is treated as first-class output, not noise.

02

Design-to-dossier

From a bare scaffold to a wet-lab-ready package: predicted structure, optimized variants, a developability profile, and a falsifiable hypothesis with explicit pass/fail acceptance criteria and a ready assay plan.

03

Confidence-tiered output

Every result is tiered — well-characterized, class-level, or novel — so you always know how much of a candidate is grounded in known biology versus advanced as a hypothesis to test.

04

Verifiable provenance

Each result ships with a verifiable provenance attestation — an integrity-and-timestamp record plus public-source citations — so your teams can confirm what was done and when, while the underlying methods stay protected.

What the process produces

Output is organized by type and labelled by confidence — never presented as experimental evidence. This is what a candidate looks like when it leaves the platform.

Output What it is Confidence
Characterized candidate A candidate whose scaffold maps to well-understood biology, with predicted structure and a worked assay plan. Grounded
Class-level candidate A candidate that aligns to a known structural class but warrants direct testing to confirm behavior. Class-level
Novel-scaffold candidate A candidate with no known analog — surfaced and labelled rather than discarded; the richest source of new chemistry. Novel · hypothesis
Design package Predicted structure, optimized variants, developability profile, falsifiable hypothesis, and assay plan — wet-lab-ready. Spec-complete
Provenance attestation Integrity-and-timestamp record plus public-source citations accompanying every result. Verifiable

Every output above is computational. “Confidence” describes how well-grounded a prediction is — not experimental confirmation. Wet-lab validation is the next-stage step.

What we don’t claim

  • We are not a wet lab. We produce computational designs and test plans, not experimental evidence.
  • We don’t claim cures or efficacy. Every prediction is a hypothesis to be tested, labelled as such.
  • We don’t present wild-type natural sequences as novel composition. Our defensible value is the method, the engineered variants, and use claims — held under separate filings.
  • “Computationally validated” is never “validated.” A statistical or structural result is a reason to run the experiment, not a substitute for it.
Reference Product

FirstRayn: a supplement designed by the engine.

FirstRayn is the first finished product designed via the OmniRayn genome engine. Patent-filed name: “Shiitake-Bridge.” A four-component formulation built around a Lentinula edodes BGC architecture identified by Discovery, paired with three established functional ingredients to bridge the in-vivo absorption pathway. DSHEA-compliant market-entry pathway.

Genomic Lead
Lentinula edodes

Shiitake mushroom · type I PKS BGC architecture · 50 Mb genome scan

Bridge Components
AHCC, Astragalus, Ceylon Cinnamon

Established functional ingredients chosen for absorption / pathway support.

Genomic Findings
52 BGC clusters · 290 domain hits

Across the 50 Mb genome scan. Architectural class only; no compound isolated.

Regulatory Path
DSHEA pathway

Functional supplement classification. No medical claims. Timeline subject to NDI notification and FDA response.

Brand bridging note. Commercial brand is FirstRayn; the corresponding intellectual-property filing is held under the prior internal name Shiitake-Bridge. All commercial assets reference the FirstRayn brand; the IP record stays at the filed name. Filing status is disclosed to qualified reviewers under NDA.

License inquiry Method details
The Foundry Exchange

A curated exchange for candidate intelligence.

Mechanism-anchored, tox-filtered, trial-whitespace-mapped molecular candidates — every listing carries a validation-state badge so the trust line is never blurred. Research intelligence, computational and pending wet-lab; no disease or efficacy claims. Engagements run under NDA.

Hypothesis

Computational transcriptomic prediction, pending wet-lab. The engine’s default output.

In-vitro validated

Mechanism-level signal confirmed in an assay panel. Not an efficacy or disease endpoint.

Lead

Advanced candidate — typically a novel-composition analog with a provisional filed. Structure under NDA.

Validation-state and confidence tier are two different axes — don’t conflate them. Validation-state (Hypothesis → In-vitro validated → Lead) describes whether a candidate has been tested in a wet lab. Confidence tier (Grounded · Class-level · Novel) describes how well-grounded the underlying computational signal is. A candidate can be Class-level in confidence yet still at Hypothesis validation-state — the biology is well-understood, but that specific candidate hasn’t been assayed.

The current catalog is at Hypothesis validation-state. In-vitro and Lead tiers populate as wet-lab data is acquired through partner co-validation.

From the catalog
Mechanism screen Hypothesis

Selective mitochondrial-function inducers — clinically untried

Real LINCS L1000 compounds ranked by cytotox-corrected selective upregulation of a mitochondrial-function gene set. Tox-clean gate applied; cross-referenced against the trials registry.

Tox-clean · trial-whitespace · mechanism-class literature cited (PubMed)
Dossier Hypothesis

Caffeic-acid phenethyl ester (CAPE) — aging-program reverser

A propolis-derived natural product flagged as a selective aging-program reverser, clinically untried. Supplement-lane eligible under structure-function framing — substantiation first, never disease claims.

Natural product · tox-clean · mechanism-class literature cited (PubMed)
Engagement tiers
Everyone

Teaser

Free

Headline whitespace counts + mechanism summaries.

Prosumer · biohacker

Digest / dossier

Pricing on inquiry

One dossier or the monthly whitespace digest, with cited literature.

Small supplement startup

Indie

Pricing on inquiry

Commercial-use catalog for one mechanism.

Mid-size biotech / brand

Pro

Pricing on inquiry

All mechanisms + custom-screen credits + early access.

Pharma scouting / funds

Enterprise

Pricing on inquiry

Everything in Pro + first-look on novel-composition leads (NDA).

Developers

License

Negotiated · under NDA

Exclusive / non-exclusive novel-composition rights. Structure under NDA; Foundry retains title.

Custom discovery

Bring us your target. We screen it.

A custom computational discovery run on an organism or compound you choose — foundation-model mining (or connectivity screen) with orthogonal null-control validation, delivered as a prioritized candidate package with falsification design. Computational, pending wet-lab; NDA-gated. Larger runs & programs quoted to scope.

From $5,000 · per single-target screen
Request a scoped screen → Or pay for a standard single-target screen · $5,000

Pricing is provided under inquiry. Novel-composition structures are NDA-gated and never shown publicly. Confidential mechanisms are not listed on public surfaces.

Browse the full exchange Request access & rate card (NDA)
The Foundry Exchange · Dossier Store

Candidate dossiers — $499 each

Browse 160 mechanism-anchored, tox-filtered, clinical-trial-whitespace-mapped candidate dossiers across 20 compounds — or whole-compound bundles ($1,499). Each is a computational hypothesis, pending wet-lab — structure-function intelligence on a known compound, not medical, efficacy, or freedom-to-operate advice. Delivered as a branded PDF to your email.

All dossiers are commercial-use licensed — $499 / dossier, $1,499 / bundle. Academic research & education use is free via the teaching packages; escalate to commercial anytime. Enterprise & catalog licensing scoped on qualification.

Before you buy. All signals are computational predictions PENDING WET-LAB VALIDATION; structure-function framing only — no disease, treatment, or efficacy claim is made. At checkout you accept the Terms of Sale / Data License and acknowledge each dossier is non-refundable once accessed. Not evaluated by the FDA.
Browse the Exchange →

160 dossiers · 20 compounds · live catalog · secure Stripe checkout · emailed within minutes

Research

Peer-review paper · public release.

The OmniRayn Discovery methodology paper documents the two-channel null-control validation framework and the biosynthetic-gene-cluster discovery results, at architectural-class level. Full per-cluster detail and methodology internals are available to qualified reviewers under NDA.

OmniRayn Discovery: A Null-Control-Validated Survey of Biosynthetic Gene Cluster Candidates

Biosynthetic gene cluster discovery has historically ranked candidates against shuffled-sequence controls alone. We document a two-channel null-control framework that adds an intergenic null — random non-BGC regions of the organism's own genome — and apply it across seven medicinal Basidiomycete and Ascomycete genomes.

Class-level empirical results: 131 candidate clusters across seven genomes, with one robust Type I polyketide candidate surviving every null control applied. Architectural-class evidence only — no compound has been isolated, characterised, or tested. Research-stage; methodology preprint forthcoming.

Authors: OmniRayn Discovery Research
Status: Peer-review paper, research-stage. Full text with methodological detail is provided to qualified reviewers under NDA — research@omnirayn.com.

Peer-review paper · 2026
A Null-Control-Validated Survey of Biosynthetic Gene Cluster Candidates
OmniRayn Discovery Research
7 medicinal-fungal genomes
131 candidate BGC loci
Two-channel null-control framework
1 robust Type I polyketide lead
BGC discovery · null-control methodology · architectural-class evidence
IP

Intellectual Property.

OmniRayn Discovery technology and assets are protected under one or more intellectual-property filings. Filing status, application numbers, and continuation strategy are not disclosed in marketing material and are available to qualified reviewers under NDA.

Trademarks: OmniRayn, OmniRayn Discovery, and FirstRayn.

Get in touch

One inbox. One brand.

All conversations run under NDA. We respond within one business day.

Research · Discovery · FirstRayn · Exchange · Reviewer Track
research@omnirayn.com

One inbox for everything — discovery and FirstRayn licensing, pharma BD & pipeline, Foundry Exchange access, and reviewer-copy requests for academic reviewers and foundation grant evaluators. Mutual NDA template available.

Confidentiality is mutual. Customer data — genomic material, peptide sequences, and pre-publication assemblies for Discovery — is handled under written NDA. Proprietary strains and unpublished candidate sets are never included in training datasets or external reference releases.