How to read a chemical class

Start with the easy half. A structural class is a shape. Every racetam is built around a pyrrolidone ring; a five-sided ring holding one nitrogen and one oxygen, usually with a small arm hanging off it. That is the whole membership requirement. Nothing about the ring says what the molecule will do once it is inside a person.

The proof of that is sitting in ordinary medicine. Brivaracetam has the racetam skeleton and the racetam name ending, and it is an epilepsy drug that works by binding a protein on synaptic vesicles; its effect on cognition is not established ^[2]. Levetiracetam, the drug it came from, is the same story. Both are racetams in exactly the sense that piracetam is, and neither is a memory drug.

A functional class is the opposite. "Ampakine" is not a shape; it means a molecule that turns up the response of the AMPA glutamate receptor without switching the receptor on by itself. Any skeleton that does that job qualifies. "Eugeroic" is looser still: it just means something that keeps you awake without behaving like a classic stimulant. The word was coined in the 1980s for adrafinil, the compound that came before modafinil, and it describes a result rather than a mechanism.

There is a quick test. Ask whether the name could be drawn. If a chemist could sketch the required skeleton from the word alone, it is structural. Then ask whether the name promises an outcome. If it does, it is functional, and it is a claim that somebody has to have earned with evidence.

Almost nothing in this field was invented from a blank page. A chemist takes a molecule that already works, keeps the part that does the work, and changes one thing. The part that stays is the scaffold or parent skeleton. The thing that changes is a substituent; a group of atoms bolted onto a specific position.

There are only a handful of reasons to make that change. To get more of the drug into the brain, by making it more fat soluble, since a greasy molecule crosses out of the blood more readily than a watery one. To make it survive longer, by blocking the exact spot where a liver enzyme would otherwise chew it up. To make it stronger, so a smaller dose does the same job. To make it more selective, so it hits one receptor and leaves its neighbours alone. Or to stretch out how long it lasts.

The clearest worked example in this whole subject is piracetam. Hang a single phenyl ring off the fourth position of its pyrrolidone ring and you get phenylpiracetam. That one ring makes the molecule far more fat soluble, and the behaviour changes character: in mice, phenylpiracetam increases locomotor activity, and the two mirror-image forms of it do not behave the same way, with only the R form improving memory performance ^[1]. In review, phenylpiracetam is described as more potent than piracetam and used for a wider range of indications ^[2]. Same skeleton, one addition, a different drug.

Fluorine does a similar job by a different route. It is small enough to sit where a hydrogen was without distorting the molecule, and a carbon-to-fluorine bond is very hard for an enzyme to break, so fluorine is often placed exactly where metabolism would have started. It also shifts potency, membrane crossing, and acidity ^[10]. Flmodafinil is modafinil with a fluorine added to each of its two rings ^[9].

The last common trick is a prodrug: attach something the body will cut off, so the molecule you swallow is not the molecule that works. About 5 to 7 percent of approved drugs worldwide are prodrugs ^[8]. Adrafinil is converted to modafinil in the body, and fladrafinil is converted to flmodafinil the same way ^[9]. Lisdexamfetamine carries an amino acid that has to be cut off before any amphetamine exists, which is what slows its onset down. How fast any of this happens is a pharmacokinetics question, not a class question.

Seven families cover most of what turns up. None of them is a prediction about effect, and two of them make that point better than any argument could.

Tryptamines are an indole (two fused rings, one carrying a nitrogen) joined by two carbons to another nitrogen. That family contains serotonin, melatonin, and psilocybin. A sleep hormone and a psychedelic, on the same skeleton. LSD is a more elaborate case: its ergoline structure has a tryptamine buried inside it. Where these land is covered in psychedelics and 5-HT2A.

Benzimidazoles make the point even harder. The skeleton shows up in omeprazole, which reduces stomach acid, in albendazole, which kills intestinal worms, and in the nitazene opioids, some of which are more potent than fentanyl ^[11]. Three completely unrelated jobs, one ring system.

Steroids are the third trap. Four fused rings does not mean anabolic. Allopregnanolone and ganaxolone are steroids that act on GABA-A receptors and are sedating; they do not build muscle and they do not touch the androgen receptor. "Steroid" is a shape, and anabolic steroids are one branch of it.

Peptides are the odd one out, because peptide is less a skeleton than a construction method: amino acids joined in a chain. What that shared construction predicts is not an effect but a practical problem, since the gut digests most peptides like food. That is why so many of them are injected or sprayed into the nose rather than swallowed. See peptides and bioregulators.

Functional labels are claims, and they vary wildly in how much they commit to. "Ampakine" is testable: either a molecule potentiates the AMPA receptor in an assay or it does not. "Adaptogen" is not, at least not in the same way; it was defined as raising non-specific resistance to stress, which is a description of an outcome rather than a target ^[12]. "Nootropic" started as a real category, for drugs said to improve the functional plasticity of the brain ^[3], and has since been stretched to cover roughly anything sold for focus.

The useful thing about a functional label is that it groups molecules with nothing structural in common, which is exactly what a structural name cannot do. GHRP-6 is a small peptide. MK-677 is not a peptide at all; it is a synthetic molecule built in a lab from a different starting point, orally active, releasing growth hormone from pituitary cells at low nanomolar concentrations and doing so in a way the original paper called mechanistically indistinguishable from GHRP-6 ^[7]. Two unrelated shapes, one job. That is what "secretagogue" buys you, and it is covered in growth hormone secretagogues.

The unhelpful thing is that a functional label is also the easiest thing in the world to put on a product page. Nobody has to prove a molecule is an ampakine before printing the word. Treat every functional label as a claim with an owner, and go looking for who made it and on what evidence.

Names come in three kinds, and each one leaks a different piece of information.

A code name is letters plus a number: TAK-653, RAD-140, SLU-PP-332, MK-677, LGD-4033. The letters are almost always the developer. TAK is Takeda, RAD is Radius Health, SLU is Saint Louis University, MK is Merck, LGD is Ligand. The number is a serial from an internal series, so a high number usually means many earlier attempts. The real inference is subtler: an official generic name gets assigned when a compound is heading somewhere, so a molecule that is still going by its lab code years later usually did not get there. TAK-653 is an AMPA receptor modulator that reached early human testing for depression; in one crossover study, 24 healthy volunteers showed a stimulant-like pattern on cognitive and eye-movement tests at 0.5 mg and 6 mg ^[4]. It is not an approved drug. RAD-140 was described in a 2010 preclinical paper and has never been approved for anything ^[5]. SLU-PP-332 has only been shown to raise endurance in mice ^[6].

A generic name is the opposite signal. Names like testolone (for RAD-140), ibutamoren (for MK-677) and tulrampator are assigned by a naming committee, and the endings are not decorative. The suffix -racetam is an official stem meaning an amide-type nootrope of the piracetam type; -relin marks peptides that trigger pituitary hormone release, which is why sermorelin, ipamorelin and tesamorelin all end that way; -tide marks peptides generally, as in semaglutide and tirzepatide. A shared ending on a generic name is a deliberate signal about family, and it is one of the few naming conventions that can be trusted.

A systematic chemical name is a set of assembly instructions. 2-(2-oxo-4-phenylpyrrolidin-1-yl)acetamide is phenylpiracetam, and every part of that string describes a piece of the structure. It tells you exactly what the molecule looks like and nothing whatsoever about what it does. Long chemical names are not a sign of danger or of sophistication; caffeine is 1,3,7-trimethylpurine-2,6-dione.

A trade name is a company's property. Provigil and Nootropil are trade names, one molecule can carry a dozen of them across different countries, and they are chosen to sound good. They carry no chemical information at all.

The most dangerous thing done with a class name is to use it as reassurance. "It is just a racetam." "It is just a peptide." "It is only a SARM, not a steroid." None of those is an argument about safety, because a shared skeleton says nothing about how much of a molecule is enough to matter.

The nitazene opioids are the cleanest demonstration available. They are all 2-benzylbenzimidazoles; the same core with one short chain changed. Changing only the length of that chain moved potency across orders of magnitude. Etonitazene activated the mu-opioid receptor at picomolar concentrations and produced opioid effects in mice at 3 to 12 micrograms per kilogram, more potent than fentanyl, while the versions with a shorter or longer chain on the same skeleton were less potent than fentanyl ^[11]. Anyone reasoning from "it is a benzimidazole" would have got that catastrophically wrong in either direction.

The same holds for the labels. "SARM" implies a gentler thing than a steroid, and YK-11 is sold under that label while carrying a steroid skeleton. "Peptide" covers semax in a nasal spray and compounds that push growth hormone around for months. "Steroid" covers testosterone and it covers sedatives. A label sits on the outside of the bottle and the molecule is inside it.

There is also everything the class cannot know about. It does not know the dose, the purity, what else ended up in the vial, how long the compound stays in the body, or what happens on the fourth week rather than the first; that last one belongs to tolerance and dependence.

The honest way to use a class is as a filing system. It tells you which questions are likely to be the important ones and where to look for prior work. It does not answer any of them. Once the class has pointed the way, the real reading starts at the individual compound: its own data, its own dose range, its own failures.