White Paper
A new deep learning neural network can be applied to accurately impute assay activities. Unlike traditional machine learning methods, this approach is trained on sparse bioactivity data as input.
Executive summary
A new deep learning neural network can be applied to accurately impute assay activities. Unlike traditional machine learning methods, this approach is trained on sparse bioactivity data as input. This type of data is typical of that found in commercial and public databases, enabling it to learn directly from correlations between activities measured in different assays. The deep learning solution was tested in two case studies on public domain data. Results showed that the neural network method significantly outperformed traditional quantitative structure-activity relationship (QSAR) models and other well-known approaches.