Molecular similarity search
Rank compounds by molecular fingerprint Tanimoto similarity to a query molecule.
How to search for similar molecules?
Enter a compound library and one query molecule as SMILES. Molecular similarity search calculates RDKit fingerprints and ranks every supplied compound by Tanimoto similarity. The results include a table and SMILES and SDF downloads, so you can compare two molecules or prioritize analogues within your own library.
- Paste one
SMILES optional_namerecord per line in the molecule input. - Enter exactly one molecule in the query input.
- Choose a fingerprint and run the search.
- Read the highest scores first and download the compounds or table.
Worked example
Input SMILES:
CCO ethanol
CCCO propanol
CCN ethylamineUse CCO as the query. Default Morgan fingerprints give these ranked rows:
Name SMILES Similarity Rank
ethanol CCO 1.0 1
propanol CCCO 0.5555555555555556 2
ethylamine CCN 0.3333333333333333 3Input
| Input | Accepted format | Requirements |
|---|---|---|
| Molecules | Pasted SMILES or .smiles, .smi, .txt upload | One SMILES followed by an optional name per line. Maximum file size: 50 MiB. |
| Query molecule | SMILES text or file | Exactly one query molecule. |
Put the SMILES first, for example CCO ethanol, rather than a header row or name-first column. Blank lines are ignored. Unnamed molecules receive generated identifiers. Invalid SMILES fail the job instead of being silently dropped.
Settings
| Setting | Default | What it does |
|---|---|---|
Fingerprint | morgan | Choose Morgan circular environments, RDKit bond-path fingerprints, or fixed MACCS structural keys. Similarity values depend on this choice. |
Morgan radius | 3 | Number of bond steps around each atom in a Morgan fingerprint. Used only with Morgan; the default radius is 3. |
Fingerprint size | 2048 | Number of bits for Morgan and RDKit fingerprints. More bits reduce collisions. MACCS always uses its fixed key set. |
Include chirality | Off | Encode stereochemistry in Morgan fingerprints. This setting does not affect RDKit topological fingerprints or MACCS keys. |
Results
| Output | Contents |
|---|---|
| Results table | Name, SMILES, Similarity, Rank. |
results.smi | Returned molecules or states as SMILES and identifiers. |
results.sdf | Returned chemical structures with row properties; 3D coordinates are not guaranteed. |
results.tsv | Tab-separated table with the same rows and values. |
report.json | Result rows, resolved settings and warnings. |
run.log | Tool, settings, result count and warnings; failures include their error. |
Downloads are available from Files. The exported records retain returned row order.
What does Tanimoto similarity measure?
For binary fingerprints, Tanimoto similarity is the number of shared set bits divided by the number set in either fingerprint. Scores run from 0 to 1. A score of 1 means identical fingerprints, which does not prove identical chemical structures: different molecules can share a bit representation.
Morgan fingerprints encode local atom environments; RDKit topological fingerprints encode bond paths; MACCS uses fixed structural keys. Scores from different fingerprints are not interchangeable. Keep settings fixed when comparing libraries. Chirality affects Morgan comparisons only when enabled.
Every submitted compound is returned, ranked by decreasing similarity. Ties retain input order. There is no hidden score cutoff and no search of PubChem or another external database.
Which chemical search should I use?
| Goal | Tool | Result |
|---|---|---|
| Rank approximate resemblance | Molecular similarity search | All input compounds with scores. |
| Require a ring or functional group | Chemical substructure search | Compounds matching a SMARTS pattern. |
| Select library representatives | Butina clustering | Cluster assignments and representative structures. |
FAQ
Can I use this as a Tanimoto similarity calculator for two molecules?
Yes. Enter one molecule in the library and the other in the query input. The single returned score is their fingerprint Tanimoto similarity.
Does high chemical similarity mean similar biological activity?
No. It compares the chosen chemical representation. Activity, selectivity and toxicity need separate experimental or predictive evidence.
Why did changing the fingerprint change the ranking?
Fingerprints encode different structural features. Morgan radius, bit-vector size and chirality can also affect the comparison.
Related tools
Butina clustering
Group molecules by fingerprint distance using RDKit Butina clustering.
Chemical substructure search
Find compounds matching a SMARTS query with optional stereochemistry matching.
Lipinski's rule of 5
Screen compounds for oral bioavailability using molecular weight, LogP, hydrogen bond donor, and acceptor limits.
ADMET-AI
Predict absorption, distribution, metabolism, excretion, and toxicity properties for compound libraries.
Admetica
Profile small-molecule ADMET properties across 22 Chemprop-based prediction models.
AF2BIND
Predict protein ligand-binding residues from AlphaFold2 pair representations
Brenk filter
Identify toxic and reactive fragments using Brenk structural alert filter.
eToxPred
Estimate toxicity risk and synthetic accessibility scores from SMILES for compound prioritization.
Lead-likeness filter
Screen for lead-like compounds using stricter molecular descriptor criteria for early optimization.
PAINS filter
Screen compounds for Pan-Assay Interference patterns using PAINS_A, PAINS_B, and PAINS_C filters.