Liquid handling with machine learning — the perfect screening combo?

In December, scientists at St. Jude Children’s Research Hospital in the US made public Combocat — a screening platform that, by combining acoustic liquid-handling protocols with machine learning, promises a faster way to test drug combinations and doses.

Combinations of drugs — whose effects are more than the sum of their parts — are required to create effective treatment regimens for many diseases, including cancers. However, according to St. Jude Children’s Research Hospital in the United States, as the number of new drugs and potential combinations has ‘exploded’, classical screening methods have been made impractical.

Combocat, an open-source screening platform

That’s why, to address this need, the hospital’s researchers — dedicated to advancing cures and means of prevention — created Combocat, a platform that combines specialised liquid-handling technology with machine learning, to enable larger combination screens and make discovering effective drug combinations easier.

“The field of drug discovery has lacked a way to deal with the sheer number of potential combinations, which would require impractical amounts of experimental materials to screen,” said senior co-corresponding author Dr Paul Geeleher from the St. Jude Department of Computational Biology.

“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,” Geeleher added.

The Nature Communications paper

In December 2025, the team detailed the platform’s capabilities in a paper titled ‘An open-source screening platform accelerates discovery of drug combinations’ — published open access in the journal Nature Communications (doi: 10.1038/s41467-025-66223-8).

About the platform, first and co-corresponding author Dr Charlie Wright from the St. Jude Department of Computational Biology said: “By tying together leading methods in machine learning and drug dispensing technology to power high-throughput combination screening, we performed experiments on a scale that was not feasible before.”

For proof of principle of the platform’s capabilities, 9045 pairs of drugs were tested by the researchers against a neuroblastoma cancer cell line. Multiple drug pairs with strong synergistic effects were uncovered by the screen, and the top findings were confirmed with additional experiments.

What the results demonstrated, according to St. Jude, is that Combocat can efficiently uncover promising drug combinations at scale.

(L) First and co-corresponding author Charlie Wright, PhD and (R) co-corresponding author Paul Geeleher, PhD, both of the St. Jude Department of Computational Biology. Image credit: St. Jude Children’s Research Hospital.

The path to scale

Miniaturised drug dispensing with machine learning was the means of achieving scale with the platform, with sonic technology enabling customised and efficient experimental layouts for drug dispensing.

“We incorporated acoustic liquid handlers that use sound waves to transfer tiny droplets of drugs very precisely,” Wright explained. “They use the exact minimum of each liquid you need, allowing for the use of far less material than conventional pin or pipette-based techniques, and increasing the number of testable combinations.”

By using one of Combocat’s two modes, ‘dense mode’, to inform its ‘sparse mode’, machine learning helps fill in the picture. In dense mode, the researchers measured every possible dose pairing for each drug combination. Alongside this dense mode, a sparse mode allows for even tighter resource management by predicting the full results from only a small fraction of the original data.

The confidence test

The sparse mode model was trained on hundreds of drug combination experiments generated with the platform’s dense mode approach; confidence in the approach coming when the researchers compared the machine learning predictions to measured values — finding them to be highly consistent.

“We created two screening ‘modes’, with something of a trade-off between them,” Geeleher said. “We optimised the sparse mode approach for scale, but it trades detail for efficiency, while we optimised the dense mode to obtain ultra-reliable measurements, which can’t scale. However, we showed that Combocat can combine them to analyse more combinations and validate them more rapidly than traditional approaches.”

The potentialities of open source

According to St. Jude, the platform continues a legacy of drug combination therapy innovation at the research hospital, providing a way to easily screen combinations not just for cancer researchers, but for any disease in need of new treatments.

“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,” Geeleher said.

“Combocat can help expedite the identification of potentially safe and effective drug combinations, which could ultimately yield useful and potentially practice-changing new drug combinations in the clinic,” Geeleher added.

The study’s other authors — Min Pan, Gregory Phelps, Jonathan Low, Duane Currier, Ankita Sanjali, Marlon Trotter, Jihye Hwang, Richard Chapple, Xueying Liu, Declan Bennett, Yinwen Zhang, Richard Lee and Taosheng Chen — were all from St. Jude, with the study having been supported by grants from National Institute of General Medical Sciences, National Cancer Institute, National Human Genome Research Institute and ALSAC, the fundraising and awareness organisation of St. Jude.

Top image credit: iStock.com/XH4D. Stock image used is for illustrative purposes only.