liveBook · Manning

Activity cliff

As explained in Build AI Drug Discovery Pipelines

An activity cliff is a pair or series of structurally similar molecules that exhibit a large difference in biological activity.

An activity cliff is a pair or series of structurally similar molecules whose biological activities differ sharply. It limits the reliability of similarity-based activity prediction: a small structural change can produce a large activity change. section 2.5.2

A Tanimoto similarity near 0.85 is only a drug-like heuristic, not a guarantee of shared activity. Assay, target, and chemical series matter, and the heuristic can fail at activity cliffs. Conservative similarity thresholds are therefore preferable when ranking close analogs where cliffs may occur. section 2.5.2

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Defines the phenomenon and its effect on similarity heuristics, but does not explain its chemical or biological mechanism.
Examples
Uses Malaria Box compounds and a Specs compound library to illustrate similarity comparisons, but gives no explicit activity-cliff molecule pair.
What it assumes
Assumes basic molecular descriptors and fingerprints, while introducing similarity metrics and their practical limitations.
What you can do after
Evaluate similarity-based activity assumptions more cautiously, especially when selecting thresholds or ranking close analogs.
How it connects
Ties activity cliffs to molecular similarity, biological activity, 2D fingerprints, Tanimoto similarity, Dice similarity, and ligand-based screening.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.