Presenter: Fabian Rauscher, Scientific Informatics Manager

In his role as Scientific Informatics Manager at Certara Germany GmbH, Dr. Fabian Rauscher is responsible for the provision and configuration of Certara enterprise solutions for customers from the pharmaceutical and life science industries in Europe. Dr. Rauscher completed his university education at the University of Stuttgart, where he first studied Technical Biology and earned his PhD in 2008 with work on computer simulations of antibiotic resistance proteins. Dr. Rauscher has more than a decade of experience with software for computer-aided drug development and computer systems for querying and processing scientific data in the early phase drug development environment.

Overview

The development of a drug takes an average of 10 years, with the discovery and preclinical stages taking three to four of those years. While the detailed work of discovery scientists is critical, advancements in deep learning AI analytics are providing exciting opportunities to accelerate tasks throughout the discovery pipeline. More than 6,000 scientists from big pharma to small biotech shorten the drug discovery design-make-test-analyze cycles with D360, Certara’s self-service data access and analytics platform. In the workshop our experts will explain how novel deep learning AI analytics capabilities are enhancing drug discovery workflows within D360. They will demonstrate concrete use cases to show how these deep learning models can be applied to augment drug discovery in D360 from property prediction to unstructured data analysis.

Key use cases covered, include:
- Shorten the drug discovery design-make-test-analyze cycle, using self-service data access and flexible data analytics and visualization allows scientists to create re-useable templates for retrieval of latest research data in a cohesive and reproducible way without the pitfalls of manual data manipulation.
- Automated Property Prediction- Predictive deep learning analytics models train on public and/or proprietary chemical structure and biological data to provide quantitative or categorical predictions that improve substance prioritization.
- Novel Structure Generation- Deep learning models within D360 can create and optimize chemical structure ideas for assessment by discovery scientists. When combined with predictive models, users can assess newly generated chemicals alongside existing product compounds to extend the range of chemotypes relevant to project goals.
- Harnessing Data from Unstructured Literature

- Traditionally, D360 has delivered insights from structured sources such as databases. With the addition of large language models, users can search and extract data from literature-based content to enhance existing datasets and inform research decisions with a broader ecosystem of data.