Prediction of Molecular Electronic Transitions Using Random Forests

Fluorescent molecules, fluorophores, play essential roles in bioimaging. Attachmentof fluorophores to proteins enables observation of the detailed structure and dynamicsof biological reactions occurring in the cell. Effective bioimaging requires fluorophoreswith high quantum yields to detect weak signals. Besides, fluorophores with variousemission frequencies are necessary to extract richer information. An essential com-putational component to discover novel functional molecules is to predict molecularproperties. Here, we present statistical machines that predict excitation energies andassociated oscillator strengths of a given molecule using a random forest algorithm. Ex-citation energies and oscillator strengths are directly related to the emission spectrumand the quantum yields of fluorophores, respectively. We discovered specific molecu-lar substructures and fragments that determine the oscillator strengths of moleculesfrom the feature importance analysis of our random forest machine. This discovery isexpected to serve as a new design principle for novel fluorophores.