ランチョンセミナー

The scale and complexity of bioinformatics research continue to grow, often exceeding the capacity of local computing environments. This talk begins with the fundamentals of cloud computing and introduces managed services designed for omics analysis, explaining how they can improve reproducibility, scalability, and cost efficiency of research workflows. We also touch on how recent advances in generative AI can be integrated into the research process — from accelerating literature review to assisting pipeline development — to enhance researchers' productivity. We present the potential of scalable infrastructure that frees researchers from computational constraints, allowing them to focus on the science itself.

The Protein Data Bank is running out of 4-character IDs. In preparation, world-wide Protein Data Bank (wwPDB) recently launched the new Beta PDB Archive for testing.

The new directory structure is organized around extended PDB IDs and will replace the current PDB archive when all 4-character PDB IDs are assigned. The Beta PDB archive contains all current PDB data and is updated weekly.

Join the session to hear about the future of PDB identifiers, the layout of the public archive, and the deprecation of legacy PDB-formatted files. Learn how you can start utilizing new resources today.

After this seminar, users will be able to:

- Understand the layout of the new PDB beta archive

- Learn how to use URLs to PDB data that will work now and, in the future.

We will allocate sufficient time of questions and answers. Appreciate your active participation.

Drug discovery research is currently being shaped by two concurrent waves: the rapid advancement of generative AI and the practical deployment of hardware for large-scale combinatorial optimization. In this seminar, Ahead Biocomputing Co., Ltd. will present its ongoing initiatives at the intersection of these two trends.

First, we will review the current state of generative AI and biological sequence language models for small molecules, antibodies, and peptides, and openly discuss how far these technologies have progressed in addressing real-world drug discovery challenges and what remains unresolved. We will then present a concept for virtual screening of billion-compound libraries by reformulating the task as a large-scale combinatorial optimization problem, rather than as a mere extension of conventional sequential docking calculations, and solving it using the quantum-inspired optimization solution SQBM+.

We welcome participants interested in the intersection of bioinformatics, chemoinformatics, and computer science. We also hope that this seminar will serve as an introduction to the jointly organized workshop on quantum computing scheduled for the afternoon.

The explosion of generative AI and agentic systems in life sciences has created urgent demand for biomedical knowledge resources that are not only comprehensive but also structured, traceable, and machine-actionable. DISGENET addresses this need as a curated genotype–phenotype–therapy infrastructure that turns heterogeneous biomedical evidence into structured, machine-readable intelligence for discovery and innovation.

At its core, DISGENET is built to be a trusted data layer for both human researchers and AI systems alike. I will present the platform's technical foundations that make this possible: a semantic data schema that enables integration of multiple data sources, an NLP pipeline for automated extraction of relationships and metadata from publications and clinical trials, a scoring framework that quantifies evidence strength, and rigorous provenance tracking that links every association back to its source—features that are increasingly critical for validating AI-generated hypotheses and reducing hallucination risk in downstream applications.

I will then showcase selected applications in genomic medicine and drug R&D, illustrating with concrete examples how structured, provenance-aware data accelerates and de-risks real-world workflows compared to unstructured alternatives.

A key focus will be DISGENET's readiness for agentic AI systems. Its MCP server and REST API, together with semantically structured outputs, enable direct querying by LLM-based agents and multi-agent pipelines, eliminating the need for extensive re-engineering. This positions DISGENET as a “plug-and-play” knowledge layer for the next generation of AI-driven biomedical research tools.

The session includes a live demonstration of the platform and its API/agent integration, followed by Q&A. Attendees will leave with a practical understanding of how curated, semantically structured knowledge graphs like DISGENET can serve as trustworthy backbones for generative AI applications in disease genomics and drug discovery.