A first-of-its-kind public standard
The Clarity Atlas
A structure-and-function molecular map of our entire apothecary — every one of 67 botanicals traced, live, to the human proteins its molecules are measured to touch. We know of no other supplement house that publishes this layer. Most make body claims with no molecular receipts; we publish the measured receipts, name where they stop, and refuse the claim. Every line is a molecule measured, in a laboratory, to interact with a human protein — nothing more, and we hold it to exactly that.
What we built, and why it is first
The whole supplement industry makes claims about the body and publishes no molecular evidence at all. We built the opposite: a receipts-grade mapof the measured layer for a full apothecary catalog — every herb’s marker molecules, the human proteins they are measured to engage, each line tracing to a real ChEMBL laboratory record, computed live from the data so the page can never drift from it. The discipline below — what we refuse to claim, and where the map stops — is not an apology. It is the standard. We state it plainly because being able to state it is the achievement.
Every number here is computed directly from the live map — never typed by hand. An edgeis a single line linking one plant to one protein. Source of every edge: ChEMBL measured human binding/activity data, restricted to human proteins, drawn only from each plant’s established marker constituents.
Start here
Read this part even if you read nothing else
Imagine the human body is full of tiny machines called proteins. They do almost everything inside you: they sense heat, carry signals, break down food, switch other machines on and off.
A molecule is a tiny chunk of a substance — one of the small bits a plant is made of. A target is one of those proteins — a protein a molecule can physically attach to, the way a key fits into a lock. When a molecule from a plant clicks into a protein, we say it bindsthat target. Binding is a measurable, physical fact: it either shows up in a lab test, or it doesn’t.
This map describes 67botanicals. For each, it lists the human proteins that the plant’s known molecules have been measured, in a laboratory, to interact with — to bind, block, or switch on. Every connection points back to a specific molecule and a specific database record. There are 1,032 such edges in total, reaching 345 different human proteins.
That is the whole thing. It is a map of what touches what — nothing more.
The limits, stated first
What the map does not say
- It does not say any herb treats, cures, prevents, or diagnoses anything.
- It does not say any herb “supports,” “boosts,” or “improves” any organ, system, or feeling.
- It does not say any herb replaces a medicine, a practitioner, or anything else.
- A molecule binding a protein in a test tube is not proof of any effect in a living person. A key fitting a lock does not tell you whether the door is one you’d ever want to open, how often, or what’s behind it.
Binding (measured) is not the same as effect (clinical). The map stops at the molecular level on purpose. Everything past that line — does it matter in a body, at what dose, for whom — is not on this map, and we do not pretend otherwise.
Two ways of knowing
A second map we are not drawing — and why it matters most
These plants were not chosen at random. They were chosen because tradition pointed at them first — because herbalists across thousands of years and every culture on earth found them worth using, naming, and passing down. That body of traditional knowledge is real evidence. It is hard-won, it is primary, and it is its own way of knowing. It does not need a binding number to be legitimate, and a binding number can neither confirm nor overturn it.
This map does not chart that tradition. It charts one narrow, separate thing: lab measurements of molecules touching proteins. The two live on different maps. When we say “no number, no edge,” we are describing the inclusion rule for this molecular map only — not a verdict that anything off this map is unproven. The older record is the reason the newer one exists at all.
Our standard — the no-number-no-edge rule
How a connection gets onto the map
We hold every line on this map to one named, repeatable standard — the no-number-no-edge rule: no edge exists unless a real laboratory measurement of a standard type is on record for it. It is the bar we built the atlas on, and the bar we invite anyone in this category to be measured against.
- 1
Start with a plant.
For example, turmeric.
- 2
Identify its marker molecules.
Its well-documented constituents — for turmeric, curcumin and its relatives.
- 3
Look those molecules up in ChEMBL.
A large public database of laboratory measurements — and read the human proteins each molecule has been measured to bind, inhibit (slow/block), or activate (switch on).
- 4
Keep the edge only if a real measurement exists.
ChEMBL must record a quantitative value of a standard type — a Ki, IC50, EC50, or Kd — with a recorded number. The rule is simply that such a measurement exists. No measurement of this kind, no edge.
Be precise about what this rule is
It is a type-and-presence filter: one of those four measurement types, with a numeric value, is on record. It is not a potency cutoff. We apply no minimum strength — no floor, no ceiling. A weak interaction qualifies for an edge exactly the same way a strong one does. So an edge says a measurement was made and recorded — it does not say the interaction is strong, and the presence of an edge tells you nothing about potency.
One methods detail, for reproducibility.We read up to 40 measured targets per marker compound (in ChEMBL target-ID order, not strength order). As of this build no compound reaches that cap, so today’s counts are not bounded by it — we disclose it only because it would shape any future rebuild.
What every edge carries. Each line names its marker compound and the database of origin — e.g. cayenne → TRPV1 → capsaicin, from ChEMBL. That is the receipt: every edge is sourced, by compound and by database. The one thing the line does not store is the numeric potency value itself — that was applied as the selection filter upstream in ChEMBL, and to read the exact number and assay you look the compound and target up in ChEMBL directly. We say this plainly rather than imply a number sits on every line.
For the careful reader
What “interact” actually means
We use bindloosely above, but the four measurement types are not the same molecular event, and we don’t want to flatten them:
Ki · Kd — affinity
How tightly a molecule physically holds a protein — true lock-and-key binding.
IC50 · EC50 — functional
An effect on the protein’s activity — inhibition or activation, often from an enzyme or cell assay.
For every one of them, a smaller number means a stronger interaction — tighter binding for Ki/Kd, less of the molecule needed to act for IC50/EC50. This map does not record which of the four backs any given edge — so read every edge as “measured to interact with (bind, inhibit, or activate),” and consult ChEMBL for which kind it is on the record.
What honest entries look like
Rich and bare, side by side
A clean example
Cayenne → TRPV1
Cayenne’s marker molecule is capsaicin, the compound that makes chili peppers hot. On this map it interacts with TRPV1— an ion-channel protein so closely tied to capsaicin that scientists nicknamed it “the capsaicin receptor.” We name the protein only by its molecular identity; whatever role it plays in the body is not a claim that cayenne does anything in a person. Capsaicin’s other measured targets here: CNR1, CNR2, CYP1A2, PTGS1, YARS1.
And, just as honestly
A bare entry is not a weak plant
7 entries have 2 measured targets or fewer — black pepper has exactly one. We show the rich entries and the bare ones with no cosmetic padding. Measurement count reflects research funding and commercial interest — nota plant’s traditional importance, and not its worth. The map reports the state of human measurement, not the state of nature.
The shape of the map
Most measured, least measured
Most measured today
- Albizia64 targets
- Strawberry54 targets
- Green Tea50 targets
- Chinese Skullcap49 targets
- Celery41 targets
Least measured (2 or fewer)
- Atlantic Irish Sea Moss1 target
- Black Pepper1 target
- Bladderwrack1 target
- Cordyceps1 target
- Feverfew1 target
- Milk Thistle1 target
- Turkey Tail2 targets
One honest note, stated once here: several of the largest counts come substantially from ubiquitous flavonoids (a common family of plant compounds — quercetin, rutin) that many plants share — not herb-specific chemistry. Roughly 4 proteins on the map are touched by a quarter or more of all botanicals. We do not present that overlap as a discovery; tested against chance, it is largely shared flavonoid chemistry, and we say so.
The map carries its own receipts
Three honest caveats
- 1
Coverage is uneven, and that is reported, not padded.
7 entries have 2 targets or fewer — the honest edge of what is measured for their marker compounds. Target engagement is also not clinical effect.
- 2
A big number is not always a plant's own signature.
As noted above under the shape of the map, some high counts lean on chemistry many plants share rather than herb-specific molecules — so read target count as research attention, not distinctiveness. We do not present cross-herb overlap as a finding.
- 3
These are distinct botanicals, de-duplicated.
The live map is the canonical list of 67 botanicals. Same-plant slug pairs were merged and one misfiled constituent removed, so the headline counts reflect the cleaned map, not a raw pre-merge file.
The point
Why we built it this way
Most herb writing skips straight to the body: it “supports” this, it’s “good for” that. We refuse that leap. We publish the one layer that is actually measured — a molecule recorded, in a lab, to interact with a protein — and we stop at the edge of what this map measures. We do not filter for strength, so an edge means “measured,” not “strong.”
That rigor is not the boring tradeoff — it isthe breakthrough. Anyone can write that an herb is “good for” you; almost no one will publish the measured layer underneath and then hold the line at exactly what it proves. This map does, and it stays useful to very different readers at once: a curious child can follow “this molecule sticks to that protein”; a pharmacologist can take any edge back to ChEMBL and pull the underlying number and source; a herbalist can read which traditional plants happen to be heavily measured and which are barely measured, as a map of research attention, never a ranking of the plants themselves.
One thing this map firmly does not do is explainany plant. Pairing a traditionally-used herb with a protein its molecule touches is tempting to read as “so that’s why it works” — and that inference is forbidden here. The tradition stands on its own evidence, and this map neither validates nor invalidates it.
A map is not the territory. This is a map of measured molecular handshakes — every one real, every one sourced to a compound and a database, and not one dressed up as a promise about your health.
Every entry, every count
The full map
All 67botanicals, ordered by measured-target count — read straight from the live map. The number is the count of distinct human proteins each plant’s marker molecules are measured to interact with. It is a measure of research attention, not of a plant’s worth.