Credit: Tiago Rodrigues/Cell Reports Physical Science
The article “Combating small molecule aggregation with machine learning” led by Tiago Rodrigues, and resulting from collaboration with researchers from Instituto de Medicina Molecular, Duke University and Insilico Medicine Taiwan, was published, on 13th September 2021, in the “Cell Reports Physical Science”.
In the present study, a deep neural network was built to detect false positive hits and potential assay nuisances among small molecules in screening campaigns. The results in the study led by the FFUL researcher show that a high percentage of small molecules is likely to aggregate at typical biological screen concentrations (small colloidally aggregating molecules – SCAMs) and that automating their identification is competitive with expert intuition, as assessed in a Turing test. The approach suggests that smart computational tools are a viable means to tackle some of the bottlenecks in the development of drug leads.
