Computational modeling of P450s for toxicity prediction

Area of Science:

• Pharmacology
• Computational Chemistry
• Toxicology

Background:

• Drug development is lengthy and expensive, with high failure rates due to issues like drug-drug interactions (DDIs) and adverse drug reactions (ADRs).
• Despite advancements like high-throughput screening (HTS), the number of novel drug entities remains limited, and some approved drugs are withdrawn due to toxicity.
• Understanding and predicting cytochrome P450 (CYP450)-mediated toxicity is crucial for safe and effective drug development.

Purpose of the Study:

• To review advancements in computational and machine learning (ML) modeling for cytochrome P450 (CYP450) toxicity prediction over the last decade.
• To highlight the utility of computational tools in identifying non-toxic drug molecules early in the development pipeline, thereby reducing economic burden.
• To cover key aspects of CYP-mediated toxicity, including specificity, inhibition, induction, and regioselectivity.

Main Methods:

• Review of computational and machine learning (ML) methodologies applied to CYP450 modeling.
• Analysis of various computational tools used for predicting drug toxicity.
• Examination of models addressing CYP450 specificity, inhibition, induction, and regioselectivity.

Main Results:

• Numerous computational methods for predicting CYP-mediated toxicity are available and widely used in computer-aided drug design (CADD).
• These predictive models show promise in identifying potential toxic molecules at early stages of drug discovery.
• The application of these models can significantly reduce the number of compounds requiring experimental screening, leading to cost savings.

Conclusions:

• Computational models for CYP-mediated toxicity are valuable tools in computer-aided drug design (CADD) and can aid in early toxicity prediction.
• These models have the potential to decrease high failure rates observed in preclinical and clinical drug development trials.
• There is a critical need for enhancing the accuracy, interpretability, and confidence of computational models used throughout the drug discovery process.