Study of proteins and nucleic acids using molecular dynamics and machine learning techniques

Trifan, Anda

Permalink

https://hdl.handle.net/2142/117647 Copy

Description

Title

Study of proteins and nucleic acids using molecular dynamics and machine learning techniques

Author(s)

Trifan, Anda

Issue Date

2022-11-21

Director of Research (if dissertation) or Advisor (if thesis)

Tajkhorshid, Emad

Doctoral Committee Chair(s)

Tajkhorshid, Emad

Committee Member(s)

• Ramanathan, Arvind
• Pogorelov, Taras
• Sligar, Steven

Department of Study

School of Molecular & Cell Bio

Discipline

Biophysics & Quant Biology

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

Molecular Dynamics

Abstract

With recent advances in computational power, molecular dynamics (MD) simulations have become an indispensable tool in studying biophysical phenomena. MD provides not only atomistic details of a system, but also temporal information. This dissertation presents several projects studied with MD simulations to understand underlying biophysical problems. We first characterize the lipid-protein interactions of a small GTPase, K-Ras, a signaling protein whose specific mutations cause it to remain in an active state persistently, driving oncogenic activity in cells. Using highly mobile membrane mimetic (HMMM) membranes with enhanced lipid diffusion, we show the globular domain (G-domain), adapts a preferred conformation on the surface of the membrane due to tethering by the hypervariable region (HVR). The HVR significantly restrains the conformations adapted by K-Ras, playing an important role in its orientation and subsequent availability to bind to downstream effectors. The mechanism of the drug remdesivir (RDV), a recently FDA-approved drug to treat COVID-19, is also studied in this dissertation. RDV is a nucleotide analog which incorporates into RNA and stalls its translocation in the RNA-dependent RNA polymerase (RdRP). We simulated the translocation using an array of non-equilibrium methods and show atomistic details of both the translocation event of RDV as well as control simulations with adenosine. Interaction energies characterize the interactions of the two nucleotides with their surroundings to help understand their distinct mechanisms. In order to achieve longer timescales, we developed an innovative framework to combine MD simulations with fluctuating finite element analysis (FFEA) to study the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) replication transcription complex (RTC). Here we bridge the gap between the two resolutions and use ML/AI methods to orchestrate the workflow and elucidate large scale conformational changes that the RTC undergoes. AI methods have been combined with MD simulations to also study the spike protein of SARS-CoV-2. We have investigated the dynamics of the spike within different environments, including its binding to the ACE2 human receptor as well as within the full virion. We have also studied the role of the spike glycans which form a shield on the surface of the spike. We show how AI accelerates probing the conformational landscape leading to decreased time to observe experimentally determined structures using computation. Finally, we present a novel workflow accelerating drug discovery efforts to address the COVID-19 pandemic by combining MD simulations, ML techniques, and free energy calculations. Taking advantage of supercomputing power, we are able to achieve extremely high throughput and identify inhibitors for important SARS-CoV-2 target proteins such as the papain-like protease, PLPro.

Graduation Semester

2022-12

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Anda Trifan

Owning Collections