Fackeldey, Konstantin
Refine
Year of publication
Document Type
- Article (41)
- ZIB-Report (13)
- In Proceedings (10)
- Research data (2)
- Book chapter (1)
- Doctoral Thesis (1)
- Habilitation (1)
Keywords
- Markov State Models (4)
- Markov chain (2)
- Meshfree (2)
- Mixed-Integer Programming (2)
- NESS (2)
- Non-reversible Markov Processes (2)
- Schur decomposition (2)
- Atomistic to Continuum (1)
- COVID-19 (1)
- Conformation Dynamics (1)
Institute
- Numerical Mathematics (61)
- Computational Molecular Design (44)
- Modeling and Simulation of Complex Processes (20)
- Parallel and Distributed Computing (6)
- Mathematical Algorithmic Intelligence (5)
- Mathematical Optimization (2)
- Mathematical Optimization Methods (2)
- Mathematics for Life and Materials Science (2)
- AI in Society, Science, and Technology (1)
- Computational Systems Biology (1)
Molecular dynamics (MD) simulations face challenging problems since
the timescales of interest often are much longer than what is possible
to simulate and even if sufficiently long simulation are possible the complex
nature of the resulting simulation data makes interpretation difficult.
Markov State Models (MSMs) help to overcome these problems by making
experimentally relevant timescales accessible via coarse grained representations
that also allows for convenient interpretation. However, standard
set-based MSMs exhibit some caveats limiting their approximation quality
and statistical significance. One of the main caveats results from the fact
that typical MD trajectories repeatedly re-cross the boundary between
the sets used to build the MSM which causes statistical bias in estimating
the transition probabilities between these sets. In this article, we present
a set-free approach to MSM building utilizing smooth overlapping ansatz
functions instead of sets and an adaptive refinement approach. This kind
of meshless discretization helps to overcome the recrossing problem and
yields an adaptive refinement procedure that allows to improve the quality
of the model while exploring state space and inserting new ansatz
functions into the MSM.
In this paper, we present a new, optimization-based method to exhibit cyclic behavior in non-reversible stochastic processes. While our method is general, it is strongly motivated by discrete simulations of ordinary differential equations representing non-reversible biological processes, in particular molecular simulations. Here, the discrete time steps of the simulation are often very small compared to the time scale of interest, i.e., of the whole process. In this setting, the detection of a global cyclic behavior of the process becomes difficult because transitions between individual states may appear almost reversible on the small time scale of the simulation. We address this difficulty using a mixed-integer programming model that allows us to compute a cycle of clusters with maximum net flow, i.e., large forward and small backward probability. For a synthetic genetic regulatory network consisting of a ring-oscillator with three genes, we show that this approach can detect the most productive overall cycle, outperforming classical spectral analysis methods. Our method applies to general non-equilibrium steady state systems such as catalytic reactions, for which the objective value computes the effectiveness of the catalyst.
In this paper, we present a new, optimization-based method to exhibit cyclic behavior in non-reversible stochastic processes. While our method is general, it is strongly motivated by discrete simulations of ordinary differential equations representing non-reversible biological processes, in particular molecular simulations. Here, the discrete time steps of the simulation are often very small compared to the time scale of interest, i.e., of the whole process. In this setting, the detection of a global cyclic behavior of the process becomes difficult because transitions between individual states may appear almost reversible on the small time scale of the simulation. We address this difficulty using a mixed-integer programming model that allows us to compute a cycle of clusters with maximum net flow, i.e., large forward and small backward probability. For a synthetic genetic regulatory network consisting of a ring-oscillator with three genes, we show that this approach can detect the most productive overall cycle, outperforming classical spectral analysis methods. Our method applies to general non-equilibrium steady state systems such as catalytic reactions, for which the objective value computes the effectiveness of the catalyst.
This paper is concerned with the design, analysis, and implementation of concurrent coupling approaches where different (atomic and continuous) models are used simultaneously within a single simulation process. Thereby, several problems or pitfalls can happen, for example, the reflection of molecular movements at the “boundary” between the atomic and continuum regions which leads to an unphysical increase in energy in the atomic model. We investigate the problems with the aim of giving an introduction into this field and preventing errors for scientists starting their research towards multiscale methods.
Spectral clustering methods are based on solving eigenvalue problems for the identification of clusters, e.g., the identification of metastable subsets of a Markov chain. Usually, real-valued eigenvectors are mandatory for this type of algorithms. The Perron Cluster Analysis (PCCA+) is a well-known spectral clustering method of Markov chains. It is applicable for reversible Markov chains, because reversibility implies a real-valued spectrum. We extend this spectral clustering method also to non-reversible Markov chains and give some illustrative examples. The main idea is to replace the eigenvalue problem by a real-valued Schur decomposition. By this extension, non-reversible Markov chains can be analyzed. Furthermore, the chains need not have a positive stationary distribution. And additionally to metastabilities, dominant cycles and sinks can be identified, too.
The rebinding effect is a phenomenon which occurs when observing a ligand-receptor binding process. On the macro scale this process comprises the Markov property. This Makovian view is spoiled when switching to the atomistic scale of a binding process. We therefore suggest a model which accurately describes the rebinding effect on the atomistic scale by allowing ''intermediate'' bound states. This allows us to define an indicator for the magnitude of rebinding and to formulate an optimization problem. The results form our examples show good agreement with data form laboratory.
Ergopeptides, like ergocornine and a-ergocryptine, exist in an S- and in an R-configuration. Kinetic experiments imply that certain configurations are preferred depending on the solvent. The experimental methods are explained in this article. Furthermore, computational methods are used to understand this configurational preference. Standard quantum chemical methods can predict the favored configurations by using minimum energy calculations on the potential energy landscape. However, the explicit role of the solvent is not revealed by this type of methods. In order to better understand its influence, classical mechanical molecular simulations are applied. It appears from our research that “folding” the ergopeptide molecules into an intermediate state (between the S- and the R-configuration) is mechanically hindered for the preferred configurations.
Medizin aus dem Computer
(2013)
We introduce a generalized operator for arbitrary stochastic processes by using a pre-kernel, which is a generalization of the Markov kernel. For deterministic processes, such an operator is already known as the Frobenius-Perron operator, which is defined for a large class of measures. For Markov processes, there exists transfer operators being only well defined for stationary measures in $L^2$. Our novel generalized transfer operator is well defined for arbitrary stochastic processes, in particular also for deterministic ones. We can show that this operator is acting on $L^1$. For stationary measures, this operator is also an endomorphism of $L^2$ and, therefore, allows for a mathematical analysis in Hilbert spaces.
Obtaining a sufficient sampling of conformational space is a common problem in molecular simulation. We present the implementation of an umbrella-like adaptive sampling approach based on function-based meshless discretization of conformational space that is compatible with state of the art molecular dynamics code and that integrates an eigenvector-based clustering approach for conformational analysis and the computation of inter-conformational transition rates. The approach is applied to three example systems, namely n-pentane, alanine dipeptide, and a small synthetic host-guest system, the latter two including explicitly modeled solvent.
It is the ultimate goal of concurrent multiscale methods to
provide computational tools that allow to simulation physical processes
with the accuracy of micro-scale and the computational speed of
macro-scale models. As a matter of fact, the efficient and scalable
implementation of concurrent multiscale methods on clusters and supercomputers
is a complicated endeavor. In this article we present the parallel
multiscale simulation tool MACI which has been designed for
efficient coupling between molecular dynamics and finite element codes.
We propose a specification for a thin yet versatile interface for the
coupling of molecular dynamics and finite element codes in a modular
fashion. Further we discuss the parallelization strategy pursued in
MACI, in particular, focusing on the parallel assembly of transfer
operators and their efficient execution.
Trajectory- or mesh-based methods for analyzing the dynamical behavior of large molecules tend to be impractical due to the curse of dimensionality - their computational cost increases exponentially with the size of the molecule. We propose a method to break the curse by a novel square root approximation of transition rates, Monte Carlo quadrature and a discretization approach based on solving linear programs. With randomly sampled points on the molecular energy landscape and randomly generated discretizations of the molecular configuration space as our initial data, we construct a matrix describing the transition rates between adjacent discretization regions. This transition rate matrix yields a Markov State Model of the molecular dynamics. We use Perron cluster analysis and coarse-graining techniques in order to identify metastable sets in configuration space and approximate the transition rates between the metastable sets. Application of our method to a simple energy landscape on a two-dimensional configuration space provides proof of concept and an example for which we compare the performance of different discretizations. We show that the computational cost of our method grows only polynomially with the size of the molecule. However, finding discretizations of higher-dimensional configuration spaces in which metastable sets can be identified remains a challenge.
The rebinding effect is a phenomenon which occurs when observing a ligand-receptor binding process.
On the macro scale this process comprises the Markov property.
This Makovian view is spoiled when switching to the atomistic scale of a binding process.
We therefore suggest a model which accurately describes the rebinding effect on the atomistic scale by allowing ''intermediate'' bound states.
This allows us to define an indicator for the magnitude of rebinding and to formulate an optimization problem.
The results form our examples show good agreement with data form laboratory.
Particle methods have become indispensible in conformation dynamics to
compute transition rates in protein folding, binding processes and
molecular design, to mention a few.
Conformation dynamics requires at a decomposition of a molecule's position
space into metastable conformations.
In this paper, we show how this decomposition
can be obtained via the design of either ``soft'' or ``hard''
molecular conformations.
We show, that the soft approach results in a larger metastabilitiy of
the decomposition and is thus more advantegous. This is illustrated
by a simulation of Alanine Dipeptide.
In contrast to the well known meshbased methods like the finite element method, meshfree methods do not rely on a mesh. However besides their great applicability, meshfree methods are rather time consuming. Thus, it seems favorable to combine both methods, by using meshfree methods only in a small part of the domain, where a mesh is disadvantageous, and a meshbased method for the rest of the domain. We motivate, that this coupling between the two simulation techniques can be considered as saddle point problem and show the stability of this coupling. Thereby a novel transfer operator is introduced, which interacts in the transition zone, where both methods coexist.