![\"Close-up](\"https:\/\/www.newsbeep.com\/us\/wp-content\/uploads\/2026\/09\/d41586-026-02882-x_53696722.jpg\"\/)
<\/p>\n <\/p>\n
Artificial-intelligence-based models of protein structure could be improved by incorporating data from pharmaceutical companies.Credit: Miyako Nakamura\/Getty<\/p>\n
For drug discovery, protein-folding models such as AlphaFold have a data problem: there isn\u2019t enough of it in public databases. Improving the performance of these artificial-intelligence-based tools will require extra data that provide examples of how proteins and drugs interact, some scientists argue.<\/p>\n
Protein structures\u2014 locked away by the thousands in drug company vaults\u2014 offer one promising source. Today, a consortium of pharmaceutical companies reports that using such data to train AI models of protein folding improves model performance markedly.<\/p>\n
The group used OpenFold3 \u2014 an open-source replication of AlphaFold 3 \u2014 to develop a new model trained on more than 20,000 proprietary protein structures. The system outperformed both comparable ones trained on public data alone and those trained on the siloed datasets of individual firms. The study, described in a blog post, has not been peer-reviewed, and the model is not publicly available.<\/p>\n
\u201cYou add all this data, and you get a pretty big bump in performance,\u201d says Mohammed AlQuraishi, a computational biologist at Columbia University in New York City, who was part of the effort.<\/p>\n
AlphaFold is running out of data \u2014 so drug firms are building their own version<\/p>\n <\/a><\/p>\n The findings, he says, strengthen the case for generating similar publicly available datasets to supercharge protein-folding AIs. One such project, called OpenBind and supported by up to \u00a38 million (US$10.8 million) in UK government funding, released hundreds of new protein structures last month, with thousands more in the works.<\/p>\n An untapped vein<\/p>\n The Protein Data Bank (PDB), an open repository of more than 200,000 experimentally determined protein structures, was the bedrock of AlphaFold 2\u2019s training data<\/a>. It enabled the tool to predict protein structures with startling accuracy \u2014 a breakthrough recognized<\/a> with the 2024 Nobel Prize in Chemistry<\/a>.<\/p>\n The model\u2019s successors, including AlphaFold 3, added the ability to predict how proteins will interact with other molecules, including potential drugs<\/a>. But the PDB has relatively few examples of experimentally determined structures interacting with drug-like molecules \u2014 maybe just 10,000, says Paul Mortenson, vice-president for computational chemistry and informatics at Astex Pharmaceuticals in Cambridge, UK.<\/p>\n That lack of data is a problem for drug-discovery efforts. Research has suggested that the accuracy of AlphaFold 3 and other \u2018co-folding\u2019 models \u2014 which predict the structure of proteins interacting with each other \u2014 drops off a cliff when the models are challenged to predict interactions between molecules highly dissimilar to those on which they were trained1<\/a>.<\/p>\n To make such tools more useful in drug discovery, researchers say, they need access to more data \u2014 which is why they have turned to vaults of molecular structures from pharmaceutical companies.<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/us\/wp-content\/uploads\/2026\/09\/d41586-026-02882-x_52147804.png\"\/)
<\/p>\n
![\"\"](\"https:\/\/www.newsbeep.com\/us\/wp-content\/uploads\/2026\/09\/d41586-026-02882-x_20323140.png\"\/)
<\/p>\n