With collaborators at SRI International we have recently demonstrated how we can use machine learning for prediction of Ultraviolet-visible (UV-Vis) spectra prediction. UV-adVISor is a new computational tool for predicting UV-Vis spectra from a molecule’s structure. We have produced two different spectrum datasets (Dataset I, N = 949 and Dataset II, N = 2222) using different compound collections and spectrum acquisition methods to train, validate, and test the models. We have evaluated the prediction accuracy of the complete spectra by the correspondence of wavelengths of absorbance maxima and with a series of statistical measures of the entire spectrum curve. This work is described in a recent publication in Analytical Chemistry.

The UV spectrum of a compound is useful for predicting important optical and chemical properties, such as phototoxicity, which is evaluated for potential drugs prior to Phase III clinical trials. Development of dyes for biotechnology, genomics, immunoassays, and drug discovery makes frequent use of the UV-Vis absorbance spectra of molecules. Predicting the UV-Vis spectrum of a compound before synthesis and experimental testing enables avoiding molecules that may interfere with high throughput assays 5 and produces other benefits in terms of cost of manufacture and speed. UV-adVISor is able to provide fast and accurate predictions for libraries of compounds. You can submit compounds here.