Arthur Ecoffet1,2, Frédéric Poitevin3, Khanh Dao Duc1,4,5. 1. Department of Mathematics, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada. 2. Department of Applied Mathematics, Ecole Polytechnique, Palaiseau, 91120, France. 3. Department of Data Analytics, LCLS, SLAC National Accelerator Laboratory, Menlo Park, CA, 94025, USA. 4. Department of Computer Science, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada. 5. Department of Zoology, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada.
Abstract
MOTIVATION: Cryogenic Electron-Microscopy offers the unique potential to capture conformational heterogeneity, by solving multiple 3 D classes that co-exist within a single cryo-EM image dataset. To investigate the extent and implications of such heterogeneity, we propose to use an optimal-transport based metric to interpolate barycenters between EM maps and produce morphing trajectories. RESULTS: While standard linear interpolation mostly fails to produce realistic transitions, our method yields continuous trajectories that displace densities to morph one map into the other, instead of blending them. AVAILABILITY: Our method is implemented as a plug-in for ChimeraX called MorphOT, which allows the use of both CPU or GPU resources. The code is publicly available on GitHub (https://github.com/kdd-ubc/MorphOT.git), with documentation containing tutorial and datasets. SUPPLEMENTARY INFORMATION: User manual for MorphOT is available at Bioinformatics online.
MOTIVATION: Cryogenic Electron-Microscopy offers the unique potential to capture conformational heterogeneity, by solving multiple 3 D classes that co-exist within a single cryo-EM image dataset. To investigate the extent and implications of such heterogeneity, we propose to use an optimal-transport based metric to interpolate barycenters between EM maps and produce morphing trajectories. RESULTS: While standard linear interpolation mostly fails to produce realistic transitions, our method yields continuous trajectories that displace densities to morph one map into the other, instead of blending them. AVAILABILITY: Our method is implemented as a plug-in for ChimeraX called MorphOT, which allows the use of both CPU or GPU resources. The code is publicly available on GitHub (https://github.com/kdd-ubc/MorphOT.git), with documentation containing tutorial and datasets. SUPPLEMENTARY INFORMATION: User manual for MorphOT is available at Bioinformatics online.