Apheris Gateway

The Apheris Gateway is a federated, end-to-end solution designed for life science organisations seeking to advance computational drug discovery through secure, privacy-preserving collaboration. It enables data providers — whether pharmaceutical companies, research institutions, or clinical partners — to maintain full control of their proprietary datasets while collaboratively training AI models across organisational boundaries. Data never moves; instead, computations travel to the data, allowing sensitive IP and regulated assets to remain protected at all times.

Built on federated learning principles and enhanced with robust governance and auditability features, the Apheris Gateway allows organisations to both build and join life science data networks. It is purpose-built for the diversity and complexity of drug discovery data, supporting every stage from target identification through to lead optimisation.

Supported Data Types and Formats

• Supports all major life science data formats, including SMILES and InChI for chemical structures, proteomics and metabolomics datasets, assay results, and medical imaging formats such as DICOM, TIFF, and OMOP.
• Any data type can be registered to the Apheris Gateway, accommodating the full diversity of drug discovery data in a single platform.

Data Control and Governance

• Data owners retain complete control over who can access their datasets and which data-driven algorithms are permitted to run.
• Organisations can define precisely how computations unfold, ensuring sensitive IP and regulated assets are never exposed.
• Data never leaves its source environment — computations visit the data, not the other way around.
• Collaboration is initiated with a simple agreement, removing the need to dismantle existing data silos or infrastructure.

Privacy-Preserving Dataset Discovery

• The Gateway enables organisations to discover and evaluate external datasets before committing to full collaboration, covering everything from chemical screening data to clinical trial insights.
• Customisable dataset details and metadata allow data providers to share meaningful context without exposing raw data.
• Synthetic discovery data and privacy-preserving data analytics packages provide a clear picture of a dataset's potential, enabling early-stage testing with confidence.

Data Preprocessing and Harmonisation

• Supports harmonisation of complex, heterogeneous datasets across multiple research sites, including molecular fingerprints (SMILES), omics matrices, clinical records, and imaging data.
• Users can run pre-built routines, custom workflows, or third-party tooling through the Gateway to unify data modalities.
• Designed to streamline translational research by reducing friction in cross-site data preparation.

Federated Model Deployment and Integration

• The Apheris Registry supports deployment of federated models across all stages of drug discovery, including well-known models such as OpenFold-3, BioNeMo, and XGBoost, as well as custom models.
• Models are deployed in a privacy-preserving manner directly from the Registry.
• A programmatic interface allows integration of custom machine learning pipelines, bespoke workflows, and specialised bioinformatics tools.
• Supports one unified network that can serve any application across the drug discovery pipeline.

Auditability and Compliance

• The Apheris Gateway generates detailed logs of all interactions with the product and with data, providing a comprehensive audit trail.
• Compliance can be easily monitored and demonstrated to auditors using built-in logging capabilities.
• Data scientists receive per-computation logs to support monitoring and efficient debugging of federated workflows.

The Apheris Gateway follows a secure-by-design architecture and offers flexible APIs and deployment options, making it straightforward to integrate into existing infrastructure. Its design allows scientific teams to focus on discovery rather than setup, while meeting the security and compliance requirements of regulated life science environments.