FQS.AI Toolkit

The FQS.AI Toolkit is a suite of software and consulting services from FQS Poland (a subsidiary of Fujitsu Limited) targeting researchers and developers in materials science and drug discovery. It applies machine learning and AI methods to tasks including data analysis, predictive modeling, and rational design of molecules and materials.

The toolkit is intended to integrate with existing research workflows and is available for both pharmaceutical and materials science applications. A trial version is available for download.

Core Features

• Advanced Machine Learning Models: Includes a range of algorithms covering deep learning architectures and ensemble methods such as Random Forest models, applicable across multiple use cases.
• Cheminformatics and Materials Informatics: Provides tools for data mining, feature engineering, and molecular or materials representation, with support for RDKit and other libraries.
• Predictive Modeling: Supports building and validating QSAR and QSPR models for predicting ADMET properties, material properties, biological activity, and related endpoints.
• Generative Models: Uses generative AI to design novel molecules and materials with specified properties, enabling exploration of chemical and materials space.
• Explainable AI (XAI): Integrated XAI tools provide insight into model decisions and the underlying structure-property relationships driving predictions.
• Customizable Workflows: Workflows can be adapted to specific research needs, and users can incorporate their own data and algorithms.
• Scientific Software Development: FQS Poland offers custom software development services to address specific research challenges.

Pharmaceutical Applications

• Drug Discovery and Design: Supports virtual screening, de novo drug design, and lead optimization for identifying and refining lead compounds.
• ADMET Prediction: Builds models for predicting absorption, distribution, metabolism, excretion, and toxicity to reduce the risk of late-stage development failures.
• Pharmacokinetics: Simulates and predicts pharmacokinetic properties of drug candidates to inform dosing and clinical planning.
• In Silico Alternatives to Animal Testing: Provides in silico models for predicting toxicity and other properties as alternatives to animal-based testing.

Materials Science Applications

• Materials Discovery: Screens materials databases and uses generative models to identify or design new materials with target properties.
• Materials Property Prediction: Predicts material properties to support optimization for specific applications.
• Process Optimization: Uses AI to identify key parameters influencing material properties and manufacturing performance.
• Computational Materials Science: Augments existing computational workflows with AI tools for data analysis, model building, and high-throughput screening.

Reported Benefits

• Accelerated R&D: Aims to reduce time and cost by automating and accelerating key research tasks.
• Improved Decision Making: Predictive models and XAI tools provide insight into structure-property relationships to support more informed decisions.
• Risk Reduction: Early identification of potential issues such as toxicity or instability through predictive modeling.
• Faster Innovation: Generative AI and high-throughput screening capabilities support exploration of a broader solution space.

Support and Services from FQS Poland

• Team expertise spans computational chemistry, cheminformatics, materials science, data science, and AI.
• Prior experience includes the scientific software product SCIGRESS.
• Works with clients to develop tailored solutions based on specific research requirements.
• Provides training services covering both computational chemistry and the application of AI in chemistry.
• Offers ongoing support to assist users in applying the toolkit effectively.

FQS Poland became a subsidiary of Fujitsu Limited in April 2021. The FQS.AI Toolkit is offered alongside consulting services, and the company positions itself as a collaborative partner rather than a software-only vendor.