Turbo similarity searching: Effect of fingerprint and dataset on virtual‐screening performance

Abstract

Abstract. Turbo similarity searching uses information about the nearest neighbours in a conventional chemical similarity search to increase the effectiveness of virtual screening, with a data fusion approach being used to combine the nearest-neighbour information. A previous paper suggested that the approach was highly effective in operation; this paper further tests the approach using a range of different databases and of structural representations. Searches were carried out on three different databases of chemical structures, using seven different types of fingerprint, as well as molecular holograms, physicochemical properties, topological indices and reduced graphs. The results show that turbo similarity searching can indeed enhance retrieval but that this is normally achieved only if the similarity search that acts as its starting point has already achieved at least some reasonable level of search effectiveness. In other cases, a modified version of TSS that uses the nearest-neighbour information for approximate machine learning can be used effectively. Whilst useful for qualitative (active/inactive) predictions of biological activity, turbo similarity searching does not appear to exhibit any predictive power when quantitative property data is available.

The Effect of Adding Indirect Relationship to Turbo Similarity Searching

“…TSS usually obtains the best result recall value with either 50NN or 100NN. NCI database did not perform well in both of the searching strategies; while MDDR and World of Molecular BioActivity (WOMBAT) provided a more higher and consistent recall [14].…”

“…Another previous work by [14] was done to determine and test the effect of different sets of structural representations (fingerprints) and database on the final performance of the chemical similarity searching strategies. Three types of fingerprints were used in the study.…”

Chemoinformatics at the University of Sheffield 2002–2014

“…The similar property principle would suggest that the nearest neighbours of the reference structure in a similarity search are likely to exhibit the same bioactivity; if we then assume that they actually do exhibit that activity then we can use them as pseudo-reference structures in a group fusion search, combining the rankings resulting from their use with that resulting from the initial reference structure. [38][39] Thus far, we have described data fusion as involving the combination of multiple rankings of a database to produce a single, fused ranking that is the output from a similarity search. Our initial studies used simple arithmetic fusion rules that had first been described for the combination of rankings in textual information retrieval systems [40] as exemplified in Table 2.…”

Clustering files of chemical structures using the Székely–Rizzo generalization of Ward's method

“…The version of WOMBAT used here contained 138,127 molecules, and searches were carried out for the fourteen activity classes described by Gardiner et al [35]. The molecules were represented by ECFC_6 fingerprints, analogous to the ECFC_4 fingerprints (i.e.…”