AI tool accelerates drug repurposing: And it’s free

Finding effective treatments for the myriad diseases that plague humanity requires innovative approaches to drug discovery. The process is extremely lengthy and expensive, taking up to 15 years and costing over a billion dollars. Now, TxGNN, a groundbreaking artificial intelligence (AI) model, has been developed that promises to improve the way treatment options can be identified for diseases with limited treatment options. It is also being made available free of charge to encourage clinicians and scientists to advance the search for new therapies.

Leveraging recent advances in geometric deep learning and human-centered AI, the Zitnik lab at Harvard has introduced TxGNN, a model designed to revolutionize drug repurposing and discovery. TxGNN is a graph neural network pre-trained on a comprehensive knowledge graph covering 17,080 clinically recognized diseases and 7,957 therapeutic candidates. This rich dataset enables TxGNN to handle various therapeutic tasks, such as predicting indications (appropriate uses for a drug) and contraindications (situations in which a drug should not be used), in a transparent format that makes drug discovery easier for researchers.

TxGNN is the first AI model specifically designed to identify drug candidates for rare diseases and conditions for which there are no treatments. It represents the largest number of diseases that a single AI model can handle to date. The researchers believe that TxGNN could be used for even more diseases.

Zero-Shot Inference: A Turning Point

One of the most notable features of TxGNN is its ability to perform zero-shot inference. This means that the model can make predictions about new diseases without requiring additional parameters or fine-tuning of the ground truth labels. In other words, TxGNN can identify potential therapeutic applications for diseases it has never encountered before, making it a powerful tool in the fight against rare and neglected diseases. According to experts, zero-shot inference is a technique in which a model can classify or predict information for a completely new category even if it has not seen any training examples for that particular category.

Combating rare and undiagnosed diseases

There are more than 7,000 rare and undiagnosed diseases worldwide. Although each disease affects only a small number of people, these diseases affect approximately 300 million people worldwide, taking a staggering human and economic toll. Because only 5 to 7 percent of these diseases have an FDA-approved drug, they remain largely untreated or undertreated. Developing new drugs for these diseases is a daunting challenge, but TxGNN offers hope and the free tool has been released specifically to facilitate more work in this area.

The performance of TxGNN was simply impressive. It could not only find drugs for specific conditions, but also predict which drugs would have side effects and contraindications, which in the current approach to drug discovery are mainly determined by trial and error during early clinical trials that focus on safety. In evaluations, the model showed significant improvements over existing methods, achieving up to 49.2% higher prediction accuracy for indications and 35.1% higher accuracy in identifying contraindications.

In the past, drug repurposing – identifying new therapeutic uses for approved drugs – was often a matter of chance. By using a geometric deep learning framework, TxGNN can make therapeutic predictions even for diseases for which there are no drugs. This capability is particularly valuable for treating complex, neglected, or rare diseases, which often lack pre-existing indications and known molecular target interactions.

Interpretable and transparent predictions

One of TxGNN's key strengths is its interpretability. The model's Explainer module provides transparent insights into the multi-hop paths that form TxGNN's predictive foundation. This feature allows clinicians and scientists to contextualize and validate the model's predictions, making it easier to trust and act on the proposed therapy candidates. In user studies, medical experts found these explanations helpful in understanding and validating TxGNN's predictions.

TxGNN identifies common disease mechanisms based on shared genomic underpinnings, allowing it to infer from a well-understood disease with known treatments to a poorly understood disease with no treatments. This ability brings the AI ​​tool closer to the kind of reasoning a human clinician might use to develop new ideas if they had access to all the pre-existing knowledge and raw data that the AI ​​model has, but that the human brain cannot possibly access or store.

The tool was trained on massive amounts of data, including DNA information, cell signals, gene activity levels, clinical notes, and more. The researchers tested and refined the model by having it perform various tasks. The tool's performance was validated using 1.2 million patient records, identifying drug candidates for various diseases and predicting patient characteristics that would make the identified drug candidates contraindicated for certain populations.

Future directions

The success of TxGNN opens up exciting new opportunities for drug discovery. The model's ability to generalize to diseases with few treatment options and perform zero-shot inference makes it a versatile tool that can be adapted for other use cases, such as drug target discovery and targeted therapy selection. In addition, the multi-disease prediction strategy used by TxGNN suggests that comprehensive drug repurposing approaches can generate more repositioned drug candidates than traditional approaches that focus on only one area.

TxGNN represents a significant advance in the field of drug discovery. By leveraging the power of geometric deep learning and human-centered AI, the model offers a promising solution to the challenge of identifying new therapeutic options for diseases with limited treatment options. With its impressive performance, real-world validation, and transparent predictions, TxGNN is poised to have a lasting impact on the way we discover and deliver drugs, ultimately improving patient care and outcomes.