RDKit Aggregator Advisor

A Knime (2.12.1) workflow that uses RDKit nodes for assessing the risk of a compound to form aggregates under biochemical assay conditions, inspired by the Aggregation Advisor paper of Irwin et al. While the Aggregator Advisor web tool allows for aggregation risk assessment of individual compounds, this Knime workflow can be used to quickly run through large compound collections.

Query structures are loaded as an SD file (here a set of 1090 approved drugs), and are then compared for Tanimoto similarity to a set of 12640 known aggregators disclosed in the Irwin paper. The compounds are then classified according to Aggregation Alert using the Rule Engine node:

High risk: Tanimoto similarity with any of the known aggregators >= 0.85 and SlogP > 3

Medium risk: Tanimoto similarity with any of the known aggregators >= 0.85 or SlogP > 3

Low risk: Tanimoto similarity with any of the known aggregators < 0.85 and SlogP < 3

Known aggregators will be identified as having Tanimoto similarity of 1.0.