We use cookies to understand how you use our site and to improve the overall user experience. This includes personalizing content and advertising. Read our
Privacy Policy
Profacgen offers ADMET Prediction service, delivering state-of-the-art computational assessment of Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) properties directly from two-dimensional chemical structures. ADME describes the disposition of a pharmaceutical compound within an organism and is applied throughout drug discovery to investigate drug levels and the kinetics of drug exposure to tissues, ultimately influencing both the performance and pharmacological activity of candidate molecules. Our service covers a broad panel of in silico assays that mirror classical experimental investigations—including membrane permeability, metabolic stability, and toxicity endpoints—enabling rapid and accurate evaluation of compounds long before they reach the bench.
By leveraging advanced ADME and toxicity prediction algorithms, our scientists calculate and interpret more than 100 properties—spanning solubility, logP, pKa, and beyond—for collections of molecules such as synthesis candidates and compound libraries. Compounds with unfavorable properties are eliminated early, while structural refinements that improve ADMET profiles are proposed prior to synthesis. Our optimized protocol further combines ADMET predictions with virtual screening (VS) and QSAR methods, delivering significant cost reduction and accelerating the path from hit to lead.
Why ADMET Prediction?
ADMET properties are among the most frequent causes of late-stage attrition in drug development. Predicting these properties rapidly and accurately from 2D structures helps research teams prioritize the right compounds, reduce synthetic effort on flawed chemotypes, and design better molecules from the very beginning of a project. Key capabilities that make computational ADMET prediction indispensable include:
Early Risk Triage: Identify and deprioritize compounds with unfavorable permeability, metabolic instability, or toxicity signals before synthesis, conserving budget and timelines.
Rational Structural Refinement: Use property predictions to guide medicinal chemistry decisions, proposing modifications that improve bioavailability, reduce clearance, or lower toxicity risk.
Broad Property Coverage: Evaluate over 100 physicochemical, ADME, and toxicity endpoints from a single 2D input, including solubility, logP, pKa (with all microstates), and key pharmacokinetic parameters.
Large-Scale Library Screening: Calculate ADMET profiles for entire compound libraries or virtual enumerations, enabling data-driven compound selection and library design.
Custom Model Development: Build proprietary QSPR/QSAR models trained on your internal datasets to capture project-specific structure–property relationships and improve prediction relevance.
Figure 1. Importance of ADMET in drug discovery.
Our ADMET Prediction Service Offerings
Service Component
Description
Physicochemical Property Prediction
Calculation of fundamental molecular descriptors including molecular weight, logP, logD, aqueous solubility, pKa with inclusion of all microstates, and polar surface area (PSA) to characterize compound developability.
Absorption & Distribution Prediction
Prediction of Caco-2 permeability, blood–brain barrier (BBB) penetration, plasma protein binding, and oral bioavailability to estimate in vivo exposure and tissue distribution.
Metabolism Prediction
Assessment of CYP450 substrate and inhibition profiles, CYP metabolite generation, and metabolic stability to anticipate first-pass effects and drug–drug interaction risk.
Excretion & Toxicity Prediction
Estimation of clearance, half-life, LD50, hERG inhibition, AMES mutagenicity, and hepatotoxicity to flag safety liabilities and guide compound deselection.
Integrated ADMET Profiling & Optimization
Multi-parameter optimization combining all ADMET endpoints, structural refinement proposals to improve profiles, and custom QSAR/QSPR model development trained on client datasets.
Comprehensive Property Panel: Over 100 ADMET and physicochemical endpoints predicted from 2D structures, including solubility, logP, pKa, and key PK parameters such as rate of metabolism and LD50.
Microstate-Aware pKa Prediction: Inclusion of all microstates in pKa calculations ensures accurate ionization profiles across physiologically relevant pH ranges.
CYP Metabolite Generation: Automated enumeration of CYP450-mediated metabolites provides mechanistic insight into metabolic fate and potential reactive intermediates.
Confidence-Estimated Models: Classification models provide confidence estimates for each prediction, allowing researchers to weigh results and prioritize follow-up experiments appropriately.
Custom QSPR/QSAR Model Building: Our scientists, with in-depth understanding of structure–property relationships, develop proprietary models trained on large, well-validated datasets—including your own data—for project-specific accuracy.
Representative Case Studies
Case 1: ADMET-Guided Lead Optimization for Poor Oral Bioavailability
Background:
A pharmaceutical company approached Profacgen with a lead series that exhibited strong target potency in vitro but suffered from poor oral bioavailability in rodent PK studies, threatening the project's progression.
Our Solution:
Our team performed integrated ADMET profiling on the lead series and close analogs. Predictions identified high first-pass metabolism via CYP3A4 and low aqueous solubility as the primary drivers of the poor bioavailability. Using multi-parameter optimization, we proposed targeted structural modifications designed to reduce CYP3A4 liability while improving solubility, all while preserving the key pharmacophore responsible for potency.
Final Results:
Guided by the ADMET predictions, the client synthesized a refined set of analogs. The optimized lead demonstrated oral bioavailability improved from 8% to 45%, with target potency maintained, enabling the program to advance to the next development milestone.
Case 2: Compound Library Toxicity Triage
Background:
A client planning a major screening campaign wanted to deprioritize compounds with significant toxicity risk before committing resources to synthesis and experimental testing of a 10,000-compound library.
Our Solution:
Profacgen screened the entire library against predictive toxicity models covering hERG inhibition, AMES mutagenicity, and hepatotoxicity. Each compound received a risk score with associated confidence estimates, and high-risk structures were flagged for deprioritization.
Final Results:
Approximately 23% of the library was flagged as high-risk and removed from the synthesis queue. The triage saved the client an estimated $500K in unnecessary synthesis and testing costs while enriching the screening set with compounds having a more favorable predicted safety profile.
Q: What input format is required for ADMET prediction?
A: We accept standard chemical structure formats including SMILES, SDF, MOL, and InChI. For library-scale projects, a single SDF or CSV file containing multiple structures can be processed in batch mode.
Q: How accurate are the ADMET predictions?
A: Our prediction models are trained using large, well-validated datasets and provide confidence estimates for classification endpoints. While in silico predictions cannot fully replace experimental assays, they are highly effective for ranking, prioritization, and guiding structural optimization, particularly when used in a comparative manner across a congeneric series.
Q: Can you build custom QSAR models using our proprietary data?
A: Yes. Creation of your own QSPR/QSAR models is a core offering. Our scientists work with your internal experimental data to develop project-specific models that capture the unique structure–property relationships of your chemical series, improving prediction relevance for novel analogs.
Q: How does pKa prediction handle ionizable compounds?
A: Our pKa prediction includes all microstates, accounting for every possible protonation and tautomeric form of a molecule. This microstate-aware approach provides an accurate ionization profile across physiologically relevant pH ranges, which is essential for reliable predictions of solubility, permeability, and distribution.
Q: Can ADMET predictions be combined with virtual screening and QSAR?
A: Absolutely. Our optimized protocol combines ADMET predictions with virtual screening (VS) and QSAR methods. This integrated workflow enables simultaneous optimization of potency and developability, leading to significant cost reduction by eliminating compounds with unfavorable properties and proposing structural refinements before synthesis.
Q: What does a typical project turnaround look like?
A: Turnaround depends on project scope. Single-compound predictions are typically delivered within a few business days, while library-scale screening of thousands of compounds may take one to two weeks. Custom QSAR model development, which involves data curation and model validation, is usually completed within two to four weeks depending on dataset size and complexity.
References:
Du BX, Xu Y, Yiu SM, Yu H, Shi JY. ADMET property prediction via multi-task graph learning under adaptive auxiliary task selection. iScience. 2023;26(11):108285. doi:10.1016/j.isci.2023.108285
Fan N, Chen J, Wang J, Chen ZS, Yang Y. Bridging data and drug development: Machine learning approaches for next-generation ADMET prediction. Drug Discovery Today. 2025;30(11):104487. doi:10.1016/j.drudis.2025.104487
Online Inquiry
Fill out this form and one of our experts will respond to you within one business day.
Figure 1. Importance of ADMET in drug discovery.