Computer Vision for Bioinformatics 

Cryo-electron tomography (cryo-ET) is a powerful method for visualizing the three-dimensional organization of supramolecular complexes, such as the cytoskeleton, within their native cellular and tissue contexts. Due to its minimal electron dose and the reconstruction artifacts arising from the missing wedge during data collection, cryo-ET typically yields noisy density maps with anisotropic XY versus Z resolution. Molecular crowding further complicates the challenge of automatically detecting various components of supramolecular complexes. This project focuses on introducing novel tracing methods for determining the precise location of actin filaments, which are important components of the cytoskeleton, in both experimental and simulated tomograms. We present three different tools for tracing diverse types of actin filaments in various experimental and simulated datasets.

Electron cryo-microscopy (cryo-EM) is a promising technique for studying the three-dimensional structure of macromolecular complexes. It complements X-ray crystallography and Nuclear Magnetic Resonance techniques. At medium resolutions, such as 5-10 Angstroms, it is extremely difficult to determine the structure directly from the volumetric data. This work focuses on developing computational methods and tools for the automatic segmentation and quality assessment of protein structures in medium-resolution cryoEM density maps.