St. Jude Unveils Machine Learning Platform to Revolutionize Drug Discovery

A new screening platform developed by scientists at St. Jude Children’s Research Hospital promises to transform the discovery of effective drug combinations. Named Combocat, this innovative tool leverages machine learning and advanced liquid handling technology, enabling researchers to explore a vast array of drug combinations more efficiently than traditional methods allow.

Many diseases, particularly various forms of cancer, necessitate the use of drug combinations that work synergistically to enhance treatment efficacy. The sheer number of potential drug combinations presents a significant challenge for researchers, making conventional screening approaches increasingly impractical. By introducing Combocat, the St. Jude team aims to address these challenges, making the discovery process both faster and more resource-efficient.

The findings surrounding Combocat were published in the journal Nature Communications. Senior co-corresponding author Paul Geeleher, Ph.D., from the St. Jude Department of Computational Biology, emphasized the platform’s resource-saving design. “We designed Combocat to use minimal resources and enable scientists to test massive numbers of drug combinations, rapidly nominating those most likely to have synergistic effects to explore further,” he stated.

Combining Technology for Enhanced Drug Screening

Combocat integrates miniaturized drug dispensing techniques with machine learning algorithms to facilitate large-scale screening. Utilizing sonic technology, the platform allows for precise transfer of tiny drug droplets, ensuring minimal waste of experimental materials. “We incorporated acoustic liquid handlers that use sound waves to transfer tiny droplets of drugs very precisely,” said Charlie Wright, Ph.D., a first and co-corresponding author.

This innovative method drastically increases the number of combinations that can be tested, addressing the limitations of traditional pin or pipette techniques. The platform operates in two modes: “dense mode,” which measures every possible dose pairing for drug combinations, and “sparse mode,” which predicts results based on a smaller subset of data. The latter allows for tighter resource management while still delivering reliable predictions.

Through rigorous testing, researchers evaluated 9,045 pairs of drugs against a neuroblastoma cancer cell line, identifying several combinations with strong synergistic effects. The effectiveness of these combinations was later confirmed through additional experiments, demonstrating Combocat’s ability to efficiently uncover promising drug pairings at scale.

A Legacy of Innovation in Drug Discovery

Combocat builds on a longstanding legacy of drug combination therapy innovation at St. Jude. This platform is not only designed for cancer researchers but also has the potential to benefit any field seeking new treatment options. “We’ve created a platform that’s free, open-source, and highly usable that could become a strong standard in the drug combination discovery field,” said Geeleher.

The implications of this research are significant. By expediting the identification of potentially safe and effective drug combinations, Combocat could lead to breakthroughs in clinical treatments. As researchers continue to test and refine this platform, the potential for practice-changing new drug combinations remains promising.

For more information or to access Combocat, interested parties can visit the platform’s dedicated webpage. The research was conducted by William C. Wright and colleagues, and the full study can be found in Nature Communications (2025).