Supervised extensions of chemography approaches: case studies of chemical liabilities assessment

Abstract

Chemical liabilities, such as adverse effects and toxicity, play a significant role in modern drug discovery process. In silico assessment of chemical liabilities is an important step aimed to reduce costs and animal testing by complementing or replacing in vitro and in vivo experiments. Herein, we propose an approach combining several classification and chemography methods to be able to predict chemical liabilities and to interpret obtained results in the context of impact of structural changes of compounds on their pharmacological profile. To our knowledge for the first time, the supervised extension of Generative Topographic Mapping is proposed as an effective new chemography method. New approach for mapping new data using supervised Isomap without re-building models from the scratch has been proposed. Two approaches for estimation of model’s applicability domain are used in our study to our knowledge for the first time in chemoinformatics. The structural alerts responsible for the negative characteristics of pharmacological profile of chemical compounds has been found as a result of model interpretation.

Progress in visual representations of chemical space

“…It was shown, for example, that distance-preserving method Isomap [67] has similar discriminative ability and visualization quality to the GTM, which preserves the topology of the source data [62,68]. Isomap takes Euclidean or custom distances between all the data points in a multi-dimensional space as input and provides coordinate vectors in a low-dimensional space that best represent the intrinsic geometry of data.…”

“…In the hierarchical GTM (HGTM), interesting regions of the full GTM plot can be subject to GTM mapping again to produce more detailed representation [61]. The supervised GTM (s-GTM) employs additional parameters to distinguish the classes better [62]. The Latent Trait Model allows one to model data sets of discrete variables [46].…”

Visualization and Analysis of Complex Reaction Data: The Case of Tautomeric Equilibria

“…The performance of the clustering in the latent space can be estimated by Γ ‐score which is normalized from 0 to 1 and can be calculated for any data set where the information about classes is available. The X,Y coordinates of the data points on the map were used to calculate the Euclidian distance.…”

Analysis of the uncharted, druglike property space by self-organizing maps

“…The majority of these studies applied novel or enhanced algorithms and presented a few representative examples of such underrepresented or very unique chemical space areas. [38][39][40] For example, Zabolotna et al concluded that a few structural elements are common in the overpopulated parts (e.g., amides and sulfonamides) but found no difference between tangible and in-stock libraries [41]. The studies used different databases (ChEMBL, PubChem [42]) and compared different even more libraries (ChEMBL, PubChem, GDB [43], etc.…”