Skip to main content

Application of machine learning to APT data to trace nanoscale protein structure


EMSL Project ID
51152

Abstract

The objective of this project is the development and application of deep learning algorithms aimed at extracting molecular structure within Atom Probe Tomography (APT) data collected from protein crystals. Specifically, the computational tools and approach we develop here will be designed to mine a 3D molecular model of a 3D protein crystal of known composition and structure to define a series of key repeating molecular units making up a representative carbon backbone. Effort will then be turned to mining experimental APT 3D data to search for, and string together the same molecular structural units to identify key protein fragments as well as the orientation of the protein. This new approach will ultimately provide the unique quantitative mapping of complex protein structure at the atomic-to-nanoscale.

Project Details

Start Date
2019-10-02
End Date
2021-03-31
Status
Closed

Team

Principal Investigator

Daniel Perea
Institution
Environmental Molecular Sciences Laboratory

Co-Investigator(s)

James Evans
Institution
Environmental Molecular Sciences Laboratory

Team Members

Graham Orren
Institution
Environmental Molecular Sciences Laboratory

Mark Wirth
Institution
Environmental Molecular Sciences Laboratory

Ty Prosa
Institution
Cameca Instruments Inc.

Trevor Moser
Institution
Environmental Molecular Sciences Laboratory

Bobbie-Jo Webb-Robertson
Institution
Pacific Northwest National Laboratory