A. James Clark School of Engineering
Permanent URI for this communityhttp://hdl.handle.net/1903/1654
The collections in this community comprise faculty research works, as well as graduate theses and dissertations.
Browse
4 results
Search Results
Item Molecular dynamics simulation and machine learning study of biological processes(2022) Ghorbani, Mahdi; Klauda, Jeffery B; Brooks, Bernard R; Chemical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)In this dissertation, I use computational techniques especially molecular dynamics (MD) and machine learning to study important biological processes. MD simulations can effectively be used to understand and investigate biologically relevant systems with lengths and timescales that are otherwise inaccessible to experimental techniques. These include but are not limited to thermodynamics and kinetics of protein folding, protein-ligand binding free energies, interaction of proteins with membranes, and designing new therapeutics for diseases with rational design strategies. The first chapter includes a detailed description of the computational methods including MD, Markov state modeling and deep learning. In the second chapter, we studied membrane active peptides using MD simulation and machine learning. Two cell penetrating peptides MPG and Hst5 were simulated in the presence of membrane. We showed that MPG enters the model membrane through its N-terminal hydrophobic residues while Hst5 remains attached to the phosphate layer. Formation of helical conformation for MPG helps its deeper insertion into membrane. Natural language processing (NLP) and deep generative modeling using a variational attention based variational autoencoder (VAE) was used to generate novel antimicrobial peptides. These in silico generated peptides have a high quality with similar physicochemical properties to real antimicrobial peptides. In the third chapter, we studied kinetics of protein folding using Markov state models and machine learning. We studied the kinetics of misfolding in β2-microglobulin using MSM analysis which gave us insights about the metastable states of β2m where the outer strands are unfolded and the hydrophobic core gets exposed to solvent and is highly amyloidogenic. In the next part of this chapter, we propose a machine learning model Gaussian mixture variational autoencoder (GMVAE) for simultaneous dimensionality reduction and clustering of MD simulations. The last part of this chapter is about a novel machine learning model GraphVAMPNet which uses graph neural networks and variational approach to markov processes for kinetic modeling of protein folding. In the last chapter, we study two membrane proteins, spike protein of SARS-COV-2 and EAG potassium channel using MD simulations. Binding free energy calculations using MMPBSA showed a higher binding affinity of receptor binding domain in SARS-COV-2 to its receptor ACE2 than SARS-COV which is one of the major reason for its higher infection rate. Hotspots of interaction were also identified at the interface. Glycans on the spike protein shield the spike from antibodies. Our MD simulation on the full length spike showed that glycan dynamics gives the spike protein an effective shield. However, breaches were found in the RBD at the open state for therapeutics using network analysis. In the last section, we study ligand binding to the PAS domain of EAG potassium channel and show that a residue Tyr71 blocks the binding pocket. Ligand binding inhibits the current through EAG channel.Item PHONON MEDIATED THERMAL TRANSPORT IN TRANSITION METAL DICHALCOGENIDES(2020) Peng, Jie; Chung, Peter W; Mechanical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Transition metal dichalcogenides (TMDCs) have attracted extensive interests due to outstanding electronic, optical, and mechanical properties, thus are highly promising in nanoelectronic device applications. However, comprehensive understanding of phonon mediated thermal transport in TMDCs is still lacking despite the important roles they play in determining the device performance. The topics requiring further explorations include the full Brillouin zone (BZ) phonons, temperature dependence of thermal properties, and structural-thermal relations of TMDCs. In determining above phonon transport characteristics, the anharmonic effect plays a central role. In this thesis, we present studies on the phonon properties of two TMDC materials, namely MoS2 and HfS2. In the first study, effect of folding on the electronic and phonon transport properties of single-layer MoS2 are investigated. The atomic structure, ground state electronic, and phonon transport properties of folded SLMoS2 as a function of wrapping length are determined. The folded structure is found to be largely insensitive to the wrapping length. The electronic band gap varies significantly as a function of the wrapping length, while the phonon properties are insensitive to the wrapping length. The possibility of modulating the gap values while keeping the thermal properties unchanged opens up new exciting avenues for further applications of MoS2. In the second study, we show that anharmonic phonon scattering in HfS2 leads to a structural phase transition. For the first time, we discover the 3R phase above 300 K. In experiments, we observe a change in the first-order temperature coefficients of A1g and Eg mode frequencies, and lattice parameters a and c at room temperature. Moreover, an anomalous phonon stiffening of A1g mode below 300 K is also observed. The first-principle simulations find a phase transition at 300 K which is characterized by a change in the stacking order from AAA to ABC. The simulations are validated by good agreements with experimental measurements on all the above temperature coefficients. By comparing DFT calculations under harmonic and anharmonic phonon approximation, we attribute the phase change to be due to phonon anharmonicity. The anomalous A1g phonon stiffening is due to decrease of the intralayer thickness of the HfS2 trailayer, as temperature increases.Item MOLECULAR SIMULATION OF ANTIMICROBIAL PEPTIDE WLBU2-MOD BINDING WITH GRAM-NEGATIVE INNER MEMBRANE MIMIC(2019) Cline, Tyler Newman; Klauda, Jeffery; Chemical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Since the discovery of Penicillin in 1928 by Sir Alexander Fleming, antibiotics have been one of the most important technologies in modern medicine. Due to the lack of novel innovative methods and the gross abuse of antibiotics both in human use and agriculture, we currently face an antibiotic resistance crisis. In the last fifty years only a handful of new class of antibiotics that target gram-positive bacteria have been introduced and, in that time, no new class of antibiotics that target gram-negative bacteria have been introduced. This thesis focuses on the molecular dynamic simulations involving the cationic α-helical antibacterial peptide, WLBU2-mod (RRWVRRVRRVWRRVVRVVRRWVRR), binding with a gram-negative bacterial inner membrane (IM) mimic composed of palmitoyloleoyl PE (POPE), palmitoyloleoyl PG (POPG), and 1,1’,2,2’-tetraoctadecenoyl CL (TOCL2) in a 7:2:1 ratio respectively. The structure of WLBU2-mod was predicted using Robetta to be either a single extended α-helical structure or a bent α-helical structure. Replica exchange with solute tempering with an improved Hamiltonian acceptance protocol (REST2) was performed on WLBU2-mod to relax the peptide to an unstructured conformation in an ii explicit aqueous solution. WLBU2-mod relaxed with REST2 consists of mainly random coil and β-sheet secondary structure which matches experimental circular dichroism (CD) results collected by Aria Salyapongse and Dr. Tristram-Nagle. Experimental CD results with the IM predicted the peptide to be structured with majority α-helical secondary structure, contrary to the unstructured results of the peptide in water. Both structured and unstructured WLBU2-mod were placed in parallel 10 Å above the IM mimic and molecular dynamics (MD) was performed to observe the binding mechanism. Simulations failed to see significant bilayer thinning or penetration into the hydrophobic core but there is strong indication that our simulations represent in intermediate state toward the final binding mechanism. In order to observe more substantial binding to the IM, future projects should consider increasing the length of the simulations and flipping the orientation of the peptide to have the hydrophobic components face inward toward the bilayer. Future projects in combination with the groundwork laid out here will hopefully provide insight into how antibacterial peptides can become the answer to the resistance crisis we face today.Item Computational Studies of Membrane Models and their Interaction with a Peripheral Protein in Yeast, and Disruption of the Water-Oil Interface by a Hydrotrope(2017) Monje-Galvan, Viviana; Klauda, Jeffery B; Chemical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Biological and non-biological interfaces were studied using all-atom molecular dynamics simulations to understand the interaction between different molecules at the atomic level. Simulation were run to analyze the dynamics and structure of cell membrane models and their interaction with a specific protein. Additionally, the effect of a small alcohol at the water-oil interface was examined as a model for amphiphilic molecules, which are relevant in chemistry and biology. Previously developed organelle-specific membrane models for yeast S. cerevisiae (Biochem. 54:6852-6861) were improved to reflect leaflet asymmetry of the trans-Golgi network (TGN) and plasma membranes. Each model was built based on experimental trends to study interleaflet coupling and lipid clustering. The (previous) symmetric endoplasmic reticulum (ER) and TGN models were further used to study the effect of sterol type in the structural properties of the membrane, and lipid-protein interactions with a lipid transport protein in yeast, Osh4. The protein’s phenylalanine loop was determined to have the strongest interaction with the bilayer among the protein’s six binding regions (BBA-Biomemb. 1858:1584-1593). The protein’s lid, the ALPS-like motif (Amphipathic Lipid Packing Sensor), was also simulated with simple (2-lipid) bilayers and with the symmetric ER and TGN models. Key residues for peptide-membrane interaction were identified based on their interaction energy, and a time scale of ~1µs determined for stable peptide binding. The interfacial dynamics between water and cyclohexane were examined in the presence of a hydrotrope - an amphiphilic molecule that reduces the interfacial tension between two liquids. Simulations were run for water-cyclohexane systems and all butanol isomers separately to understand the effect of this hydrotrope’s chemical structure on the interface. The results reproduced experimental data trends, showing that a hydrotrope concentration of as little as 0.6mol% in the aqueous phase reduces the interfacial tension to nearly half the value of a binary water-cyclohexane mixture. Tert-butanol was further compared with experimental studies showing that at low concentrations (< 10mol%) the simulations accurately reproduce experimental data. In addition, theoretical correlations from simulation data show the system follows van der Waals theory of smooth interfaces, and describe the crossover behavior of this hydrotrope from surfactant-like to co-solvent based on its concentration in solution, and describe the crossover behavior of this hydrotrope from surfactant-like to co-solvent based on its concentration in solution.