A probabilistic method to report predictions from a human liver microsomes stability QSAR model: a practical tool for drug discovery

Abstract

Using data from the in vitro liver microsomes metabolic stability assay, we have developed QSAR models to predict in vitro human clearance. Models were trained using in house high-throughput assay data reported as the predicted human hepatic clearance by liver microsomes or pCLh. Machine learning regression methods were used to generate the models. Model output for a given molecule was reported as its probability of being metabolically stable, thus allowing for synthesis prioritization based on this prediction. Use of probability, instead of the regression value or categories, has been found to be an efficient way for both reporting and assessing predictions. Model performance is evaluated using prospective validation. These models have been integrated into a number of desktop tools, and the models are routinely used to prioritize the synthesis of compounds. We discuss two therapeutic projects at Genentech that exemplify the benefits of a probabilistic approach in applying the models. A three-year retrospective analysis of measured liver microsomes stability data on all registered compounds at Genentech reveals that the use of these models has resulted in an improved metabolic stability profile of synthesized compounds.

chemalot and chemalot_knime: Command line programs as workflow tools for drug discovery

“…The command line programs and KNIME framework are used at Genentech to create, validate and apply QSPR models for properties important to lead optimization such as metabolic stability [42], permeability [43] and solubility. Here, we present an example KNIME workflow to show how command line programs are used to construct a QSPR model to predict solubility (Fig.…”

A review on machine learning approaches and trends in drug discovery

“… The ECFP algorithm can be adapted to generate different types of circular fingerprints, optimized for different uses. They are among the most popular similarity search tools in drug discovery [33] , [34] , [35] , [36] , [37] , [38] and are used effectively in a wide variety of applications. They can store information about the environments surrounding each atom in a molecule and in addition to the search for similarities, ECFPs are well suited to recognizing the presence or absence of particular substructures, so they are often used in the construction of QSAR and QSPR models.…”

Metabolic stability studies of lead compounds supported by separation techniques and chemometrics analysis

“…Hence, developing a model using SVM approach has proved its value numerous times among studies, whose main goal was to create a reliable model able to predict metabolic stability values (e.g. intrinsic clearance [82] or in vitro biological half‐life [23]). Classification approach provides a great tool for researchers willing to divide their compounds into two groups (e.g.…”

“…metabolically stable and unstable compounds, according to the custom cut‐off values of in vitro half‐life or intrinsic clearance). After dividing the compounds into two groups, developed probabilistic model, instead of providing a more or less accurate value, may return a simple information whether tested compounds belong to the group of stable or unstable compounds (which is often more significant than finding a certain value for metabolic stability) [82,83]. Such models can also predict an isoform of CYP3A4 enzyme that mainly contributes toward studied derivative biotransformation.…”