Comparing Molecular Patterns Using the Example of SMARTS: Applications and Filter Collection Analysis

In a recent work, an algorithm to compare chemical patterns, written for example in SMARTS, was presented. This algorithm, called SMARTScompare, is able to assess the identity, subset relation, and similarity of a pair of patterns. Here we used an implementation of SMARTScompare to analyze SMARTS filter sets that were published in the context of, for example, high-throughput screening. We found that the difference in intentions with which the filter sets were designed is mirrored in the similarity values we calculated. The analysis revealed which patterns from one filter set are covered by filters from another set. In one case it became obvious that a filter set is more or less completely covered by another. Furthermore, we analyzed pattern hierarchies for consistency, and we propose a method to remove redundant patterns. SMARTScompare together with SMARTScompareView equips users with powerful methods to visualize, compare, and focus their filter sets.