Artificial Intelligence: Unexpected Results

Artificial intelligence (AI) is on the rise. Until now, AI applications generally have "black box" character: How AI arrives at its results remains hidden. Prof. Dr. Jürgen Bajorath, a cheminformatics scientist at the University of Bonn, and his team have developed a method that reveals how certain AI applications work in pharmaceutical research. The results are unexpected: the AI programs largely remembered known data and hardly learned specific chemical interactions when predicting drug potency. The results have now been published in Nature Machine Intelligence.

Which drug molecule is most effective? Researchers are feverishly searching for efficient active substances to combat diseases. These compounds often dock onto protein, which usually are enzymes or receptors that trigger a specific chain of physiological actions. In some cases, certain molecules are also intended to block undesirable reactions in the body - such as an excessive inflammatory response. Given the abundance of available chemical compounds, at a first glance this research is like searching for a needle in a haystack. Drug discovery therefore attempts to use scientific models to predict which molecules will best dock to the respective target protein and bind strongly. These potential drug candidates are then investigated in more detail in experimental studies.

Since the advance of AI, drug discovery research has also been increasingly using machine learning applications. As one "Graph neural networks" (GNNs) provide one of several opportunities for such applications. They are adapted to predict, for example, how strongly a certain molecule binds to a target protein. To this end, GNN models are trained with graphs that represent complexes formed between proteins and chemical compounds (ligands). Graphs generally consist of nodes representing objects and edges representing relationship between nodes. In graph representations of protein-ligand complexes, edges connect only protein or ligand nodes, representing their structures, respectively, or protein and ligand nodes, representing specific protein-ligand interactions.

"How GNNs arrive at their predictions is like a black box we can't glimpse into," says Prof. Dr. Jürgen Bajorath. The chemoinformatics researcher from the LIMES Institute at the University of Bonn, the Bonn-Aachen International Center for Information Technology (B-IT) and the Lamarr Institute for Machine Learning and Artificial Intelligence in Bonn, together with colleagues from Sapienza University in Rome, has analyzed in detail whether graph neural networks actually learn protein-ligand interactions to predict how strongly an active substance binds to a target protein.

How do the AI applications work?

The researchers analyzed a total of six different GNN architectures using their specially developed "EdgeSHAPer" method and a conceptually different methodology for comparison. These computer programs "screen" whether the GNNs learn the most important interactions between a compound and a protein and thereby predict the potency of the ligand, as intended and anticipated by researchers - or whether AI arrives at the predictions in other ways. "The GNNs are very dependent on the data they are trained with," says the first author of the study, PhD candidate Andrea Mastropietro from Sapienza University in Rome, who conducted a part of his doctoral research in Prof. Bajorath's group in Bonn.

The scientists trained the six GNNs with graphs extracted from structures of protein-ligand complexes, for which the mode of action and binding strength of the compounds to their target proteins was already known from experiments. The trained GNNs were then tested on other complexes. The subsequent EdgeSHAPer analysis then made it possible to understand how the GNNs generated apparently promising predictions.

"If the GNNs do what they are expected to, they need to learn the interactions between the compound and target protein and the predictions should be determined by prioritizing specific interactions," explains Prof. Bajorath. According to the research team's analyses, however, the six GNNs essentially failed to do so. Most GNNs only learned a few protein-drug interactions and mainly focused on the ligands. Bajorath: "To predict the binding strength of a molecule to a target protein, the models mainly 'remembered' chemically similar molecules that they encountered during training and their binding data, regardless of the target protein. These learned chemical similarities then essentially determined the predictions."

According to the scientists, this is largely reminiscent of the "Clever Hans effect". This effect refers to a horse that could apparently count. How often Hans tapped his hoof was supposed to indicate the result of a calculation. As it turned out later, however, the horse was not able to calculate at all, but deduced expected results from nuances in the facial expressions and gestures of his companion.

What do these findings mean for drug discovery research? "It is generally not tenable that GNNs learn chemical interactions between active substances and proteins," says the cheminformatics scientist. Their predictions are largely overrated because forecasts of equivalent quality can be made using chemical knowledge and simpler methods. However, the research also offers opportunities of AI. Two of the GNN examined models displayed a clear tendency to learn more interactions when the potency of test compounds increased. "It's worth taking a closer look here," says Bajorath. Perhaps these GNNs could be further improved in the desired direction through modified representations and training techniques. However, the assumption that physical quantities can be learned on the basis of molecular graphs should generally be treated with caution. "AI is not black magic," says Bajorath.

Even more light into the darkness of AI

In fact, he sees the previous open access publication of EdgeSHAPer and other specially developed analysis tools as promising approaches to shed light on the black box of AI models. His team's approach currently focuses on GNNs and new "chemical language models". "The development of methods for explaining predictions of complex models is an important area of AI research. There are also approaches for other network architectures such as language models that help to better understand how machine learning arrives at its results," says Bajorath. He expects that exciting things will soon also happen in the field of "Explainable AI" at the Lamarr Institute, where he is a PI and Chair of AI in the Life Sciences.

Mastropietro A, Pasculli G, Bajorath J.
Learning characteristics of graph neural networks predicting protein-ligand affinities.
Nat Mach Intell, 2023. doi: 10.1038/s42256-023-00756-9

Most Popular Now

DMEA 2025: Digital Health Worldwide in B…

8 - 10 April 2025, Berlin, Germany. From the AI Act, to the potential of the European Health Data Space, to the power of patient data in Scandinavia - DMEA 2025...

Is AI in Medicine Playing Fair?

As artificial intelligence (AI) rapidly integrates into health care, a new study by researchers at the Icahn School of Medicine at Mount Sinai reveals that all generative AI models may...

New System for the Early Detection of Au…

A team from the Human-Tech Institute-Universitat Politècnica de València has developed a new system for the early detection of Autism Spectrum Disorder (ASD) using virtual reality and artificial intelligence. The...

Generative AI's Diagnostic Capabili…

The use of generative AI for diagnostics has attracted attention in the medical field and many research papers have been published on this topic. However, because the evaluation criteria were...

Diagnoses and Treatment Recommendations …

A new study led by Prof. Dan Zeltzer, a digital health expert from the Berglas School of Economics at Tel Aviv University, compared the quality of diagnostic and treatment recommendations...

AI Tool can Track Effectiveness of Multi…

A new artificial intelligence (AI) tool that can help interpret and assess how well treatments are working for patients with multiple sclerosis (MS) has been developed by UCL researchers. AI uses...

Dr Jason Broch Joins the Highland Market…

The Highland Marketing advisory board has welcomed a new member - Dr Jason Broch, a GP and director with a strong track record in the NHS and IT-enabled transformation. Dr Broch...

Surrey and Sussex Healthcare NHS Trust g…

Surrey and Sussex Healthcare NHS Trust has marked an important milestone in connecting busy radiologists across large parts of South East England, following the successful go live of Sectra's enterprise...

AI-Driven Smart Devices to Transform Hea…

AI-powered, internet-connected medical devices have the potential to revolutionise healthcare by enabling early disease detection, real-time patient monitoring, and personalised treatments, a new study suggests. They are already saving lives...

Multi-Resistance in Bacteria Predicted b…

An AI model trained on large amounts of genetic data can predict whether bacteria will become antibiotic-resistant. The new study shows that antibiotic resistance is more easily transmitted between genetically...

DMEA 2025 Ends with Record Attendance an…

8 - 10 April 2025, Berlin, Germany. DMEA 2025 came to a successful close with record attendance and an impressive program. 20,500 participants attended Europe's leading digital health event over the...

A Novel AI-Based Method Reveals How Cell…

Researchers from Tel Aviv University have developed an innovative method that can help to understand better how cells behave in changing biological environments, such as those found within a cancerous...