Developing new drugs faster and cheaper with AI

Lamarr Institute, b-it and TüCAD2 form collaborative alliance for the academic development of drugs using Artificial Intelligence

High costs, numerous failures and decades of development: drug discovery and development is a complex process. Pharmaceutical research sees great potential in the application of Artificial Intelligence (AI). The Lamarr Institute for Machine Learning and Artificial Intelligence, the Bonn-Aachen International Centre for Information Technology (b-it) and the Tübingen Centre for Academic Drug Discovery (TüCAD2) at the University of Tübingen have now entered into a cooperation agreement. The collaboration focuses on the discovery of active agents that can treat malfunctions in protein kinases and thus the cause of various diseases. The scientists at the Lamarr Institute and b-it will contribute their many years of expertise in data analysis and the development of computer methods for drug research, while TüCAD2, as a leading center for academic drug research, will be responsible for medicinal chemistry and pharmacology.

Bonn/Tübingen. The path to developing new drugs is long, complex, and expensive. Of thousands of new potential active ingredients that are discovered and tested in laboratories, only a small percentage make it to clinical trials. With luck, a substance will pass all the tests and — provided it meets the high legal safety requirements — will be launched on the market as a new drug. It can take 10 to 15 years before a drug is launched on the market, with costs often running into the billions. The immense expenditure on research and development represents a major challenge for the pharmaceutical industry.

A new era of drug discovery

Artificial intelligence can help here and accelerate the development of drugs. The scientists in the Life Sciences division at the Lamarr Institute for Machine Learning and Artificial Intelligence and Life Science Informatics at the Bonn-Aachen International Center for Information Technology (b-it) are
leaders in the field of data analysis and Machine Learning (ML) for drug research. In a new collaboration with the Tübingen Center for Academic Drug Discovery (TüCAD2) at the University of Tübingen, the AI experts now want to take drug discovery of protein kinases to a new level. “Advances in AI-supported drug development promise new conceptual possibilities for improved and accelerated drug development,” says Prof. Dr. Jürgen Bajorath, Principal Investigator and Area Chair Life Sciences at the Lamarr Institute and Professor at b-it. “In this initiative, two renowned partners from the fields of drug development and AI are joining forces to shape a new era of academic drug research and development”.

The team headed by Prof. Bajorath is contributing its expertise in the development of computational methods for the discovery of protein kinase drugs to the collaboration. The discovery of kinase active substances is also a focus of the researchers at TüCAD2. Kinases are enzymes and perform the task of regulating cellular signaling pathways in the body. Protein kinases in particular play a role in signal transmission and control of various cellular processes. If these enzymes do not function properly, serious diseases such as cancer, neurological disorders or autoimmune diseases can develop.
This makes protein kinase compounds a promising target for drug research. As a leading center for academic drug research and development in Germany, TüCAD2 already has an outstanding track record: Under the leadership of Prof. Dr. Stefan A. Laufer, a total of five drug candidates have been brought to first application in humans since its foundation in 2012. “These research and development activities in Tübingen and Bonn are therefore highly complementary and represent a unique opportunity for an alliance between the two leading academic centers,” says Laufer.

Triangular AI crucial for acceptance and quality

From the search for potential drug candidates and more effective drug molecules to safety assessments and the performance of clinical trials – Artificial Intelligence can support drug research and make it more efficient in almost all phases. However, it is particularly important in fields of Life Sciences such as medicine and pharmacy that the functionalities behind the Machine Learning processes are transparent and understandable for everyone. This is why the researchers at the Lamarr Institute and b-it are focusing on “Explainable AI”, which is not only trained with bioscience data, but also uses additional knowledge and contextual information from various life science fields. “Why does Artificial Intelligence make a certain prediction? If we are to exploit the potential of AI in the Life Sciences, it must be understandable to an interdisciplinary audience. Otherwise, its use and acceptance will not go beyond theory,” says Bajorath. “In addition, the concept of triangular AI – combining data with a specific context and interdisciplinary knowledge – is crucial for the quality of the predictions.”

While data analysis and Machine Learning take place at the Lamarr Institute and b-it in Bonn, drug synthesis, pharmacology and biological tests are carried out at TüCAD2 in Tübingen. The scientists use the TüKIC library, currently the largest academic collection of protein kinase inhibitors (PKI)
with approx. 12,000 PKIs and 1 million activity data, as well as a collection of approx. 156,000 PKIs from public sources, which is curated at the Lamarr Institute and currently covers more than 80 percent of all human kinases, as a data basis.

About the partners