Internal Test Sets Studies in a Group of Antimalarials

Abstract

Topological indices have been applied to build QSAR models for a set of 20 antimalarial cyclic peroxy cetals. In order to evaluate the reliability of the proposed linear models leave-n-out and Internal Test Sets (ITS) approaches have been considered. The proposed procedure resulted in a robust and consensued prediction equation and here it is shown why it is superior to the employed standard cross-validation algorithms involving multilinear regression models.

Similarity-based Virtual Screening to Find Antituberculosis Agents based on Novel Scaffolds

“…On the other hand, extra-mechanistic virtual screening methodologies have demonstrated their ability to model the presence of activity within structurally heterogeneous groups of compounds in different therapeutic areas [68][69][70][71][72][73][74][75][76][77] as well as in predicting toxicological properties [78,79] and drug-like character [80]. In these models, structural similarity is the key.…”

Internal Test Set (Its) Method: A New Cross-Validation Technique to Assess the Predictive Capability of Qsar Models. Application to a Benchmark Set of Steroids

“…The ITS algorithm is simple and general and has been described elsewhere. [23][24][25][26] As shown below, it is very easy to extend this approach to the LMO formalism but, for the sake of simplicity, it will be described within the paradigmatic framework of a LOO protocol. The algorithm, which must be applied on a molecular set, proceeds as follows: For each molecule, the prediction made in step 2.3 is kept as a definitive result.…”

Superposing Significant Interaction Rules (SSIR) Method: A simple Procedure for Rapid Ranking of Congeneric Compounds

“…Our code is always based on the Internal Test Sets (ITS) paradigm. [17,[25][26][27][28] This means that every cross-validated training to be done over a subset must start from scratch and the variable selection must be totally independent of the other cross-validation loops or subsequent prediction over a test or validation set. This is accomplished by the above algorithm because the training and the model application are done in distinct molecular and mutually exclusive sets.…”