Semeta

Semeta, developed by Optibrium, is a metabolite prediction software purpose-built for DMPK scientists. It delivers accurate insights into Phase I and II drug metabolism by predicting metabolic routes, sites, products, and lability, enabling researchers to make well-supported decisions about which assays and preclinical experiments to run and how to interpret the results. By anticipating unexpected metabolism early, Semeta helps teams avoid costly late-stage failures and streamline their DMPK workflows.

Underpinned by quantum mechanics and machine learning models, Semeta offers greater precision and sensitivity than other available methods. Its models are grounded in a fundamental mechanistic understanding of each enzyme family, covering the full breadth of human Phase I and II metabolism as well as key preclinical species.

Key Challenges Semeta Addresses

• Predicting in vivo metabolite profiles with higher precision than competing software, supporting the interpretation of metabolite-ID experiments.
• Complementing in vitro and in vivo DMPK studies with comprehensive in silico modelling to identify enzymes, isoforms, sites of metabolism, and likely metabolites.
• Identifying compounds with multiple routes of clearance to reduce the risk of drug-drug interactions or complications arising from genetic polymorphisms.
• Choosing the most appropriate preclinical species for testing by comparing cytochrome P450 regioselectivity between human and animal species.
• Guiding compound design by identifying metabolically vulnerable sites and balancing properties using a Probabilistic Scoring approach.

Enzyme Families and Species Coverage

• Cytochrome P450s (CYPs)
• Flavin-containing monooxygenases (FMOs)
• Aldehyde oxidases (AOXs)
• Uridine diphosphate glucuronosyltransferases (UGTs)
• Sulfotransferases (SULTs)

In addition to human enzyme models, Semeta includes models for rat, dog, and mouse CYPs, enabling direct comparison of sites of metabolism across common preclinical species to inform the selection of the most appropriate animal model for experimental studies.

Interactive Visualisation Features

• Generate metabolic pathways in Semeta's Card View®, clearly displaying relationships between parent compounds and metabolites.
• Identify sites vulnerable to metabolism using regioselectivity maps labelled by enzyme and likelihood of metabolism.
• Quickly understand which enzymes and isoforms are responsible for a compound's metabolism via WhichEnzyme™ and WhichP450™ pie charts.
• Guide optimisation of CYP-mediated metabolic stability using the metabolic landscape visualisation tool.

Scientific Foundations and Peer-Reviewed Publications

• A paper in Xenobiotica (November 2023) describes a new method for determining the most likely experimentally-observed routes of metabolism and metabolites, based on the WhichP450™, regioselectivity, and WhichEnzyme™ models.
• A publication from June 2020 covers methods for modelling FMOs and UGTs to predict reactivity to drug metabolism.
• A May 2023 paper describes a model for predicting whether a particular molecular site will be metabolised by cytosolic sulfotransferase enzymes (SULTs).

Deployment and Integration

• Semeta is cloud-based, accessible from any internet-connected computer, with hassle-free deployment and easy maintenance.
• Hosted on Amazon Web Services (AWS), it is supported by world-leading security models and covered by Optibrium's ISO 27001-accredited information security system.
• Semeta's models are also available as a Metabolism module within Optibrium's StarDrop drug discovery platform, providing access to additional capabilities such as SAR analysis, generative chemistry, 3D drug design features, and broader ADME modelling.