Matching Items (2)
Filtering by
- All Subjects: enhanced sampling
- Creators: Thorpe, Michael F
Description
In a typical living cell, millions to billions of proteins—nanomachines that fluctuate and cycle among many conformational states—convert available free energy into mechanochemical work. A fundamental goal of biophysics is to ascertain how 3D protein structures encode specific functions, such as catalyzing chemical reactions or transporting nutrients into a cell. Protein dynamics span femtosecond timescales (i.e., covalent bond oscillations) to large conformational transition timescales in, and beyond, the millisecond regime (e.g., glucose transport across a phospholipid bilayer). Actual transition events are fast but rare, occurring orders of magnitude faster than typical metastable equilibrium waiting times. Equilibrium molecular dynamics (EqMD) can capture atomistic detail and solute-solvent interactions, but even microseconds of sampling attainable nowadays still falls orders of magnitude short of transition timescales, especially for large systems, rendering observations of such "rare events" difficult or effectively impossible.
Advanced path-sampling methods exploit reduced physical models or biasing to produce plausible transitions while balancing accuracy and efficiency, but quantifying their accuracy relative to other numerical and experimental data has been challenging. Indeed, new horizons in elucidating protein function necessitate that present methodologies be revised to more seamlessly and quantitatively integrate a spectrum of methods, both numerical and experimental. In this dissertation, experimental and computational methods are put into perspective using the enzyme adenylate kinase (AdK) as an illustrative example. We introduce Path Similarity Analysis (PSA)—an integrative computational framework developed to quantify transition path similarity. PSA not only reliably distinguished AdK transitions by the originating method, but also traced pathway differences between two methods back to charge-charge interactions (neglected by the stereochemical model, but not the all-atom force field) in several conserved salt bridges. Cryo-electron microscopy maps of the transporter Bor1p are directly incorporated into EqMD simulations using MD flexible fitting to produce viable structural models and infer a plausible transport mechanism. Conforming to the theme of integration, a short compendium of an exploratory project—developing a hybrid atomistic-continuum method—is presented, including initial results and a novel fluctuating hydrodynamics model and corresponding numerical code.
Advanced path-sampling methods exploit reduced physical models or biasing to produce plausible transitions while balancing accuracy and efficiency, but quantifying their accuracy relative to other numerical and experimental data has been challenging. Indeed, new horizons in elucidating protein function necessitate that present methodologies be revised to more seamlessly and quantitatively integrate a spectrum of methods, both numerical and experimental. In this dissertation, experimental and computational methods are put into perspective using the enzyme adenylate kinase (AdK) as an illustrative example. We introduce Path Similarity Analysis (PSA)—an integrative computational framework developed to quantify transition path similarity. PSA not only reliably distinguished AdK transitions by the originating method, but also traced pathway differences between two methods back to charge-charge interactions (neglected by the stereochemical model, but not the all-atom force field) in several conserved salt bridges. Cryo-electron microscopy maps of the transporter Bor1p are directly incorporated into EqMD simulations using MD flexible fitting to produce viable structural models and infer a plausible transport mechanism. Conforming to the theme of integration, a short compendium of an exploratory project—developing a hybrid atomistic-continuum method—is presented, including initial results and a novel fluctuating hydrodynamics model and corresponding numerical code.
ContributorsSeyler, Sean L (Author) / Beckstein, Oliver (Thesis advisor) / Chamberlin, Ralph (Committee member) / Matyushov, Dmitry (Committee member) / Thorpe, Michael F (Committee member) / Vaiana, Sara (Committee member) / Arizona State University (Publisher)
Created2017
Description
The structure of glass has been the subject of many studies, however some
details remained to be resolved. With the advancement of microscopic
imaging techniques and the successful synthesis of two-dimensional materials,
images of two-dimensional glasses (bilayers of silica) are now available,
confirming that this glass structure closely follows the continuous random
network model. These images provide complete in-plane structural information
such as ring correlations, and intermediate range order and with computer
refinement contain indirect information such as angular distributions, and
tilting.
This dissertation reports the first work that integrates the actual atomic
coordinates obtained from such images with structural refinement to enhance
the extracted information from the experimental data.
The correlations in the ring structure of silica bilayers are studied
and it is shown that short-range and intermediate-range order exist in such networks.
Special boundary conditions for finite experimental samples are designed so atoms
in the bulk sense they are part of an infinite network.
It is shown that bilayers consist of two identical layers separated by a
symmetry plane and the tilted tetrahedra, two examples of
added value through the structural refinement.
Finally, the low-temperature properties of glasses in two dimensions
are studied. This dissertation presents a new approach to find possible
two-level systems in silica bilayers employing the tools of rigidity theory
in isostatic systems.
details remained to be resolved. With the advancement of microscopic
imaging techniques and the successful synthesis of two-dimensional materials,
images of two-dimensional glasses (bilayers of silica) are now available,
confirming that this glass structure closely follows the continuous random
network model. These images provide complete in-plane structural information
such as ring correlations, and intermediate range order and with computer
refinement contain indirect information such as angular distributions, and
tilting.
This dissertation reports the first work that integrates the actual atomic
coordinates obtained from such images with structural refinement to enhance
the extracted information from the experimental data.
The correlations in the ring structure of silica bilayers are studied
and it is shown that short-range and intermediate-range order exist in such networks.
Special boundary conditions for finite experimental samples are designed so atoms
in the bulk sense they are part of an infinite network.
It is shown that bilayers consist of two identical layers separated by a
symmetry plane and the tilted tetrahedra, two examples of
added value through the structural refinement.
Finally, the low-temperature properties of glasses in two dimensions
are studied. This dissertation presents a new approach to find possible
two-level systems in silica bilayers employing the tools of rigidity theory
in isostatic systems.
ContributorsSadjadi, Seyed Mahdi (Author) / Thorpe, Michael F (Thesis advisor) / Beckstein, Oliver (Committee member) / Schmidt, Kevin E (Committee member) / Treacy, Michael Mj (Committee member) / Arizona State University (Publisher)
Created2018