CHEMPASS; A Tool for Chemical Structure Similarity Scoring and Clustering.

Tab 1: Chemical Similarity (single list)

Upload your list of compounds in .csv/.txt format either as CID/DTXSIDs/SMILES, select your choice of fingerprint generation, set the clustering threshold, and explore cluster-level statistics and heatmaps.

zip file for download includes a .pdf of clusters with similarity scores, molecular properties of compounds, high resolution images of heatmap and NMDS plots.

Tab 2: Compare two lists

Upload a reference list of SMILES + a target list of SMILES and for each compound in the target list, get the most structurally similar compound from the reference list.

  • Chemical Similarity (CS)
  • Compare two lists
Use example data to test the app without uploading a file.

Processing an arbitrary list of 15 CIDs provided as example data

Download example data
Upload File: [?]

Please upload a list of compounds in text (.txt) or CSV (.csv) format containing one or more of these identifiers.

  • CID: Compound Identification Number (PubChem). eg: 1983 (for Acetaminophen)
  • SMILES: Simplified Molecular Input Line Entry System, a line notation for describing the structure of chemical species. eg: CC(=O)NC1=CC=C(O)C=C1
  • DTXSID: DSSTox Substance Identifier used in the Environmental Protection Agency CompTox Dashboard. eg: DTXSID0020232
ID Name
DTXSID5025659 compound 1
DTXSID5025811 compound 2
6950 compound 3
8295 compound 4
C(C(=O)NCC(=O)O)N compound 5
CC(=O)NCCCCN compound 6

Choose either 1a or 1b for further processing

More information for 1a: [?]

-Molecular properties are calculated from SMILES using RDKit

-Input can only be a list of SMILES and/or Names of compounds

-If Names not provided by user; Names follow input order; Compound 1 = SMILES_1, Compound 2 = SMILES_2.

More information for 1b: [?]

This is a slower option; needs to connect with PubChem API

Molecules need to exist in PubChem for this option to work

Molecular properties are downloaded from PubChem

Select fingerprint type: [?]

What are Morgan fingerprints?

Morgan fingerprints encode molecular structures by considering circular atom neighborhoods.

  • ECFP4: Focuses on atom and bond connectivity (radius 2).
  • FCFP4: Focuses on functional classes of atoms like donor/acceptor roles.

Clustering Cutoff: [?]

What does Clustering Cutoff mean?

This value sets the distance threshold.

• A lower cutoff (e.g., 0.2) forms tighter, smaller clusters.

• A higher cutoff (e.g., 0.6) forms larger, looser clusters.

Adjust the cutoff depending on how strict you want your clustering to be.

For chemical structure clustering, a cutoff of 0.2–0.4 is often a good starting point.

For smaller datasets with fewer than 20 compounds, you may need to increase the cutoff beyond 0.4

Download reference data Download target data

Please upload a reference and target list of SMILES.

You can obtain SMILES for your list of compounds from the CS tab --> option 1a --> Download zip

Upload two lists of SMILES: [?]
  • Reference File: List of SMILES
  • Target File: Compounds (as SMILES) that you want compared to reference list
  • Output: For each compound in the target list, the output pdf will display the most structurally similar compound from the reference list.

Reference List:

C(C(=O)NCC(=O)O)N
CC(=O)NCCCCN

Target List:

C(CO)NCCO
C1CC(NC1)C(=O)O

If duplicates are found between reference and target list. Only distinct entries will be kept.

Select fingerprint type: [?]

What are Morgan fingerprints?

Morgan fingerprints encode molecular structures by considering circular atom neighborhoods.

  • ECFP4: Focuses on atom and bond connectivity (radius 2).
  • FCFP4: Focuses on functional classes of atoms like donor/acceptor roles.

  • Heatmap (CS)
  • Silhouette scores (CS)
  • Butina Clusters (CS)
  • NMDS Plot (CS)
  • Two list compare
Loading...
Loading...
Loading...
Loading...
Loading...