Project Grant 2151678

This three-year Project Grant from the National Science Foundation's Mathematical and Physical Sciences program (CFDA 47.049) provides $164,000 to Purdue University to develop integrated moment-based descriptors and deep neural networks for screening three-dimensional biological data. The researchers will extend mathematical moments to encompass fractional-order moment descriptors, providing a more accurate representation of 3D images. They will integrate the new moment-based approach into a deep neural network to achieve high accuracy and efficiency in classifying 3D data. A publicly available biomolecular 3D image web server will be implemented, screening protein ligand-binding pockets, functional sites, and drug molecules. Protein structures will be represented with voxel grids. The new techniques are expected to apply broadly to 3D data from various imaging disciplines and substantially influence machine learning domains. The funding also supports multidisciplinary coursework and research experiences for students at Purdue University and Saint Joseph's University.