MaNGA: a novel multi-niche multi-objective genetic algorithm for QSAR modelling

Abstract

Summary Quantitative structure–activity relationship (QSAR) modelling is currently used in multiple fields to relate structural properties of compounds to their biological activities. This technique is also used for drug design purposes with the aim of predicting parameters that determine drug behaviour. To this end, a sophisticated process, involving various analytical steps concatenated in series, is employed to identify and fine-tune the optimal set of predictors from a large dataset of molecular descriptors (MDs). The search of the optimal model requires to optimize multiple objectives at the same time, as the aim is to obtain the minimal set of features that maximizes the goodness of fit and the applicability domain (AD). Hence, a multi-objective optimization strategy, improving multiple parameters in parallel, can be applied. Here we propose a new multi-niche multi-objective genetic algorithm that simultaneously enables stable feature selection as well as obtaining robust and validated regression models with maximized AD. We benchmarked our method on two simulated datasets. Moreover, we analyzed an aquatic acute toxicity dataset and compared the performances of single- and multi-objective fitness functions on different regression models. Our results show that our multi-objective algorithm is a valid alternative to classical QSAR modelling strategy, for continuous response values, since it automatically finds the model with the best compromise between statistical robustness, predictive performance, widest AD, and the smallest number of MDs. Availability and implementation The python implementation of MaNGA is available at https://github.com/Greco-Lab/MaNGA. Supplementary information Supplementary data are available at Bioinformatics online.

Density of Deep Eutectic Solvents: The Path Forward Cheminformatics-Driven Reliable Predictions for Mixtures

“…To critically examine the predictivity of models MO59 and CO15, we compared their Williams plots [ 35 , 36 ], as presented in Figure 4 . As expected, model CO15 had a larger number (129 with h * = 0.0399) of structural outliers as compared to model MO59 (25 with h * = 0.0533).…”

“…To do so, we built the so-called Williams plot, in which standardized residuals were plotted against leverage values. Doing so permitted us to identify response and structural outliers [ 35 , 36 ]. All plots shown in the present work were conceived with Matplotlib [ 37 ].…”

Manually curated transcriptomics data collection for toxicogenomic assessment of engineered nanomaterials

“…Due to the homogenized preprocessing and manual curation of the metadata, this collection is a relevant resource for identification of toxicity biomarkers. This can be addressed by using multiple feature selection approaches 35,36 or more advanced data modelling techniques [37][38][39] . Biomarkers could also be detected by means of gene co-expression network analysis, under the assumption that central network genes play a key role in the adaptation to the exposure 40,41 .…”

Transcriptomics in Toxicogenomics, Part III: Data Modelling for Risk Assessment

“…Recently, a new methodology for feature selection from complex data, called MaNGA has been proposed [114]. MaNGA uses a multi-objective optimization strategy to identify the minimum set of predictive features with the widest AD, better predictivity capability and high stability.…”