Chemical Database

What This Is

This is a searchable database of chemical compounds pulled from multiple sources — regulatory lists, Wikipedia, PsychonautWiki, TripSit, and various government agencies (IARC, EPA, EU REACH, NTP, and others). Each chemical has been looked up on PubChem to get its SMILES notation, then run through EtoxPred to estimate toxicity.

What the Scores Mean

Tox-score (0 to 1): This comes from EtoxPred, a machine learning model trained on FDA-approved drugs (labeled safe) and TOXNET/T3DB chemicals (labeled toxic). It uses Morgan fingerprints — basically a mathematical representation of molecular structure — fed into an Extra Trees classifier. A score near 0 means the molecule looks structurally similar to known safe drugs. A score near 1 means it resembles known toxic compounds. The model gets about 80% accuracy on independent test sets, so treat it as a rough guide, not gospel.

Things to keep in mind:

• The model was trained on general toxicity, not specific endpoints like "will this give you cancer" or "will this mess with your hormones"
• A low Tox-score doesn't mean something is safe to consume. Plenty of FDA-approved drugs are dangerous at the wrong dose
• A high Tox-score doesn't mean something will definitely hurt you. The model has blind spots
• Natural products tend to score higher (0.5+) even when they're harmless, because the training data skews toward synthetic drug-like molecules

SAscore (Synthetic Accessibility, ~0 to 1 after transformation): How easy or hard the molecule would be to synthesize in a lab. Lower values mean easier to make. This is based on fragment analysis — common molecular fragments that show up a lot in known synthetic compounds get better scores. Unusual ring systems, lots of stereocenters, and weird functional groups push the score up.

Why this matters:

• Easy-to-synthesize compounds (low SAscore) are more likely to actually be available
• Hard-to-synthesize compounds are more expensive and rare

Categories

Chemicals can appear in multiple categories:

• Nootropic: From Wikipedia nootropic lists, Examine.com, and scientific literature
• Carcinogen: From IARC Group 1/2A lists, NTP Report on Carcinogens, EU CLP regulation, FDA lists, California Prop 65
• Endocrine Disruptor: From EU REACH SVHC list, EPA EDSP, TEDX list, Wikipedia
• Psychoactive: From PsychonautWiki substance index, TripSit factsheets, Wikipedia drug lists

A chemical appearing in the carcinogen list doesn't necessarily mean it causes cancer in humans at normal exposure levels — IARC classifications are hazard-based, not risk-based. Similarly, something listed as an endocrine disruptor means it has the potential to interfere with hormonal systems, but dose and exposure route matter enormously.

Reading the Triangle Graph

Each chemical gets a small triangle (CEP) chart showing three independent scores at a glance — Carcinogenic (top, red), Endocrine disruptor (bottom-left, gold), and Psychoactive (bottom-right, blue). Each score ranges from 0 to 1.

The outer triangle is the reference boundary (all scores at 1.0). The inner shaded polygon shows the compound's actual scores. Each vertex is pulled from the center toward its corner proportional to the score:

• A vertex sitting at the center means that score is 0 — no signal
• A vertex touching the outer edge means that score is 1 — maximum signal
• A tiny inner triangle near the center → low scores across the board
• A large inner triangle filling the outline → high scores across the board
• A lopsided shape pulled toward one corner → that particular axis dominates

The exact values are shown below the graph as C:0.xx E:0.xx P:0.xx. These scores come from sub-models trained on category-specific datasets, so they are independent of the overall Tox-score. A compound can have a low Tox-score but still show a high score on one of these axes, or vice versa.

How to Use This

1. Search for a chemical you're curious about
2. Check which categories it falls into — if something you're considering taking shows up as both a nootropic AND a carcinogen, that's worth knowing
3. Look at the Tox-score as a rough indicator — above 0.6, dig deeper before putting it in your body
4. Cross-reference with actual research. This database is a starting point, not medical advice
5. The SMILES string can be pasted into PubChem to find the full compound page with safety data, research papers, and more