Simulations Plus

Earlier & Faster: Combining ML and HT-PBPK to Accelerate Early Small Molecule Drug Discovery

Bassani D, Lawless M — Wed, 01 Apr 2026 12:59:22 +0000

Abstract

Accurately predicting the pharmacokinetics (PK) of small-molecule candidates early in discovery can accelerate optimization cycles, reduce animal testing, and improve the quality of compounds advancing toward the clinic. However, conventional physiologically based pharmacokinetic (PBPK) modeling is often limited by throughput and by the need for extensive in vitro inputs. In this webinar, Dr. Davide Bassani, Computational DMPK Leader at Roche, presents an evaluation of SwiftPK, a corporate high-throughput PBPK (HT-PBPK) application that enables rapid PBPK simulations at scale using machine learning (ML)–predicted ADME inputs derived from chemical structure alone. Using a large in vivo rodent PK dataset (9,000 compounds), his team at Roche assessed SwiftPK performance across ten PK endpoints. Overall, most endpoints were predicted within a three- to four-fold error range, with absolute average fold errors (AAFEs) spanning 2.90–4.15 across the full dataset. Their research further demonstrated that predictive performance improves when (i) filtering for compounds predicted to be primarily cleared by hepatic metabolism (Extended Clearance Classification System, ECCS class 2) and (ii) restricting to cases where ML input predictions carry high confidence. Dr. Bassani walks through these results, and how they highlight the successful applicability of HT-PBPK in early-phase projects, especially for ECCS2-predicted compounds and with reliable input-property projections, and illustrate how HT-PBPK can support compound ranking and decision-making when experimental data are limited or unavailable.

Pharmacokinetic Evaluation of Etoricoxib 120mg Tablets in Healthy Human Pakistani Volunteers: In-Vivo In-Silico Bridging for Bioequivalence

Masood SH, Muhammad IN, Siddiqui F, Saleem MT, Raza ML — Mon, 30 Mar 2026 11:00:21 +0000

Abstract

Background: Etoricoxib is a selective cyclooxygenase-2 inhibitor widely used for the treatment of pain and inflammatory conditions. However, comparative pharmacokinetic and bioequivalence data for locally manufactured etoricoxib formulations in the Pakistani population remain limited.

Objective: This study aimed to evaluate the pharmacokinetics and bioequivalence of a locally manufactured etoricoxib tablet compared with the innovator product in healthy Pakistani volunteers, with supportive assessment using physiologically based pharmacokinetic (PBPK) modeling.

Methods: Comparative in-vitro dissolution studies were conducted in buffer media of pH 1.2, 4.5 and 6.8 and evaluated using similarity factor (f₂) analysis. A randomized, open-label, two-treatment, two-period crossover bioequivalence study was performed in healthy male volunteers under fasting conditions. Subjects received a single oral dose of 120 mg Etoricoxib (ETO) as either the test product (Etoxib®) or the reference product (Arcoxia®), with a 14-day washout period. Plasma concentrations were quantified using a validated HPLC-UV method and pharmacokinetic parameters were estimated using non-compartmental analysis. PBPK modeling of the test product was conducted using GastroPlus® to assess model predictability.

Results: The test and reference products exhibited similar dissolution profiles across all media, with f₂ values indicating similarity. The geometric mean ratios (90% confidence intervals) of the test to reference product for C_max, AUC_0-t and AUC_0-∞ were 0.946 (0.8855-1.0135), 0.923 (0.8705-0.9795) and 0.960 (0.8955-1.0255), respectively, all within the regulatory bioequivalence acceptance range. PBPK model predictions for key pharmacokinetic parameters were within an acceptable fold-error range (≤2) compared with observed data.

Conclusion: The study demonstrated comparable systemic exposure and confirmed bioequivalence between the locally manufactured and innovator etoricoxib formulations in the Pakistani population. PBPK modeling provided supportive evidence of formulation similarity and model adequacy.

By Syed Hakim Masood, Iyad Naeem Muhammad, Fahad Siddiqui, Muhammad Talha Saleem, Muhammad Liaquat Raza

Simulations Plus Announces Strategic Collaboration Programs for AI-Enabled Modeling

Thu, 26 Mar 2026 08:26:21 +0000

Simulations Plus, Inc. (Nasdaq: SLP) (“Simulations Plus” or the “Company”), a global leader in model-informed and AI-accelerated drug development that advances biopharma innovation, today announced strategic collaboration programs with three large pharmaceutical companies to advance artificial intelligence (AI) workflows across the drug development lifecycle.

These programs apply AI within scientifically grounded modeling workflows and define how next-generation workflows are deployed at scale. The close collaboration between Simulations Plus and leading pharmaceutical organizations will provide direct insight into how AI will be integrated into real-world environments—informing product direction, workflow standardization, and future commercial models. The programs will utilize Simulations Plus’ major software platforms, including GastroPlus®, MonolixSuite™, ADMET Predictor®, and Thales™.

“Our approach to AI is grounded in how it operates within a complete system, not as a standalone capability,” said Jonathan Chauvin, Co-Chief Product & Technology Officer at Simulations Plus. “These collaborations will allow us to work alongside our partners, leveraging real-time scientific feedback and company data to continuously refine how workflows are orchestrated across our tools, ensuring AI-driven efficiencies translate into reproducible, traceable outcomes. The insights we gain will directly shape how we evolve our platform and deliver value at scale.”

Participating companies will integrate the Company’s internally developed AI agents directly into model-informed drug development (MIDD) workflows, enabling natural language interaction, automation of data processing, coordination of simulations across multiple modeling engines, and generation of interpretable outputs from complex, multi-step pipelines.

Importantly, the collaborations will also serve as a foundation for broader enterprise adoption, including direct alignment with information technology teams to define how AI-enabled capabilities are deployed, governed, and integrated within existing systems. This includes defining standards together for transparency, reproducibility, and governance as AI becomes more deeply embedded in drug development processes.

“As highlighted at our Investor Day presentation in January, AI will only fulfill its potential in drug development when it is delivered responsibly, grounded in validated science, and integrated into real workflows,” said Shawn O’Connor, Chief Executive Officer of Simulations Plus. “Our customers are choosing to work with us because of the strength of our validated scientific engines and depth of our teams who apply them daily within real workflows, enabling us to translate AI into practical, deployable solutions. These strategic collaboration programs represent an important step in moving us and our partners beyond experimentation and into practical implementation as we advance our software and services into a unified modeling ecosystem.”

Companies interested in learning about using AI-enabled workflows in their modeling can request additional information.

In Silico Designing of Palbociclib Loaded Plga Long-Acting Intramuscular Injection for Palliative Therapy of HR+/HER2− Metastatic Breast Cancer

Suryawanshi R, Joseph A, Malayandi R — Mon, 23 Mar 2026 11:08:15 +0000

Abstract

Objectives

Palbociclib (PBB) is an oral cyclin-dependent kinase 4/6 (CDK4/6) inhibitor approved for the treatment of HR+/HER2− breast cancer. However, poor adherence and limited tolerability of oral administration often compromise its therapeutic effectiveness, especially in palliative care for metastatic conditions. Dose reductions are frequently required to manage toxicity, but lower doses can still provide effective tumour control with reduced neutropenia risk, thereby improving quality of life and progression-free survival. Developing a long-acting injectable (LAI) formulation of PBB offers significant advantages for sustained therapy in advanced-stage cancer management.

Methods

This study focuses on dose selection, release profile optimisation, and the design of sterile PBB-loaded PLGA microsphere suspensions for intramuscular (IM) administration using physiologically based pharmacokinetic (PBPK) modelling and simulations. The PBPK model, developed and validated with data from oral and intravenous routes, enabled the prediction of IM pharmacokinetics. Clinical target product profiles were defined based on IC₅₀ and minimum steady-state concentration (Css, _min).

Key findings

The dose optimisation study revealed that the rational selection of dose for both strengths, with an optimised sustained-release profile (Target 2, showing minimal initial burst and controlled release reaching ~55% by day 10.5 and ~90% by day 21), achieved the desired clinical quality target product profiles.

Conclusions

The developed model will further support polymer selection, specification setting, and drug-to-polymer ratio. Incorporating PBB’s physicochemical properties and host response helps guide rational formulation design.

By Rutuja Suryawanshi , Anumol Joseph , Rajkumar Malayandi

ADMET Predictor® Tutorial Series: Deployment and Licensing

Jamois E — Fri, 20 Mar 2026 11:39:50 +0000

Master the deployment of ADMET Predictor® across your organization by understanding its flexible concurrent licensing model. This guide explains that simply opening software sessions or performing data analysis tasks, such as viewing histograms, does not consume a license. A single concurrent license allows for unlimited sessions and high-speed, multi-threaded calculations while using only one license feature from the pool. Users can manually manage these assets via the License Status panel, using “Viewer Mode” to return licenses to the network pool when finished. Additionally, the video highlights the “dual-headed” workflow possible with MedChem Designer™, which uses a transient license for small-scale property and metabolite predictions, ensuring that batch-processing tasks and individual compound analysis can run in parallel without interference. This pool remains accessible across the GUI, Command Line (Windows/Linux), and third-party platforms like KNIME or Pipeline Pilot.

ADMET Predictor® Tutorial Series: 3D Functionality Part 2

Miller D — Fri, 20 Mar 2026 11:34:40 +0000

Advance your drug discovery workflow with the 3D Virtual Screening tools in ADMET Predictor®. This session demonstrates how to move beyond single conformer generation to create high-throughput 3D databases from vendor catalogs. Key features covered include:

Database Creation: Configuring diversity, energy ranges, and torsion angle searches to store millions of 3D conformers.
GPU-Accelerated Search: Utilizing NVIDIA GPUs to screen over 7.5 million conformers in roughly 10 seconds.
Hybrid Similarity Scoring: Balancing volume overlap (atom-centered Gaussians) with custom pharmacophore features like H-bond donors, acceptors, and rings.
Advanced Filtering: Using SMARTS-based excluded motifs to filter out obvious hits and discover novel chemical scaffolds.
Alignment Visualization: Exporting results to external viewers like BIOVIA Discovery Studio to graphically inspect 3D molecular overlays

ADMET Predictor® Tutorial Series: 3D Functionality Part 1

Miller D — Fri, 20 Mar 2026 11:12:03 +0000

Master the 3D Functionality available in ADMET Predictor®. David Miller, VP of Cheminformatics, explains that these features—including 3D conformer generation and virtual screening—are included free for any user with a property prediction license. The video demonstrates:

3D Model Requirements: Why some models, such as MDCK permeability and aqueous solubility, require 3D descriptors.
Conformer Generation:distance geometry and energy minimization to create accurate 3D coordinates from 2D SD files.
Quality & Energy Assessment: How the “3D Quality” and “Energy” columns alert users to structural issues or close van der Waals contacts.
External Visualization: Setting up third-party tools like Avogadro to view generated conformers directly from the ADMET Predictor interface.
CLI Accessibility: Confirmation that all 3D features are also available via the command line for Windows and Linux.

ADMET Predictor® Tutorial Series: Python Module 1

Bachorz RA — Fri, 20 Mar 2026 10:42:33 +0000

Discover the synergy between ADMET Predictor® and the Python programming language. This session guides you through the installation of the PI-ADMET Predictor module and its dependencies like RDKit, Pandas, and Scikit-Learn. Key use cases include:

SMILES Standardization: Converting raw SMILES into cleaned, standardized forms using command-line scripts.
Property Prediction: Accessing ADMET properties asynchronously through the RESTful API.
Exploratory Data Analysis: Performing dimensionality reduction using t-SNE to visualize chemical space.
Pharmacokinetics (HTPK): Extracting and plotting CP time curves directly in Jupyter Notebooks.
3D Geometry & Metabolites: Generating 3D molecular structures and visualizing metabolite trees within your Python scripts

ADMET Predictor® Tutorial Series: Command Line (Including Linux) Part 2

Lawless M — Fri, 20 Mar 2026 10:26:16 +0000

Explore the advanced capabilities of the ADMET Predictor® command line interface (CLI). This tutorial covers the configuration of HTPK simulation modules, demonstrating how to export pharmacokinetic parameters like $C_{max}$ and $T_{max}$ for human, rat, and mouse species. Learn to use the -PRM flag to audit all simulation inputs and the -out flag to organize batch results into specific directories. A significant portion of the video is dedicated to Linux environments, detailng the extraction of the .tar.gz installation files, obtaining a Composite ID for Flexera licensing, and configuring the run_ap.sh shell script to handle multi-threaded runs in a server environment. Finally, the video showcases metabolite generation through script-file workflows, including the new feature for custom naming conventions of metabolic products.

ADMET Predictor® Tutorial Series: Command Line (Including Linux) Part 1

Lawless M — Fri, 20 Mar 2026 10:22:22 +0000

Unlock the efficiency of the ADMET Predictor® command line on both Windows and Linux platforms. This video demonstrates how to execute property predictions without the GUI, using SMILES, SDF, and XTK files. Key technical coverage includes configuring multi-threading with the -n option for faster batch processing, specifying pH-dependent solubility with -P, and generating Out of Scope (OTS) and Uncertainty (UNC) files for model validation. The tutorial goes into depth on the -W workflow flag, showing how to tailor output for different professional roles—such as predicting dominant ionization states at specific pH levels for docking studies or displaying full pKa microstate distributions for physical chemistry analysis.