Typical machine learning methods used to analyze experimental imaging data rely on tens or hundreds of thousands of training images. But Daniël Pelt and James Sethian of Berkeley Lab’s Center for Advanced Mathematics for Energy Research Applications (CAMERA) have developed what they call a “Mixed-Scale Dense Convolution Neural Network” (MS-D) that “learns” much more quickly from a remarkably small training set. One promising application of MS-D is in understanding the internal structure and morphology of biological cells to identify, for example, differences between healthy and diseased cells. In one such project in Carolyn Larabell’s lab, the method needed data from just seven cells to determine the cell structure.
A cross-disciplinary effort by Berkeley Lab scientists has yielded a new algorithmic approach for determining 3D molecular structures from single-particle X-ray free-electron laser (XFEL) imaging. Peter Zwart of the Molecular Biophysics & Integrated Bioimaging Division (MBIB) worked with James Sethian and Jeffrey Donatelli of the Computational Research Division’s Mathematics Group to create the multi-tiered iterative phasing (M-TIP) framework, which uses advanced mathematical techniques to extract nano-scale biological structures from sparse and noisy diffraction data. A paper detailing the approach was published in the Proceedings of the National Academy of Sciences.