Plenary, Keynote, Invited and Industry Speakers
His scientific work has focussed on theoretical chemistry of biomolecules, primarily through force field simulations of proteins and nucleic acids. Key aspects include the development of protein force fields, implicit solvent models, and computer codes for carrying out practical simulations. Applications have mainly been to biomolecular NMR and crystallography.
Title of Talk: MD simulations of total X-ray scattering in protein crystals
Abstract: The adaption of graphical processing units (GPUs) to biomolecular simulations has made microsecond-scale simulations of biomolecular crystals available on a nearly-routine basis. Typically, a super-cell consisting of several crystallographic unit cells becomes the periodically-repeating unit in the simulation. We have carried out multi-microsecond simulations of crystals of lysozyme (in three space groups), insulin, thaumatin, apoferritin and myoglobin. Here are some ways this data might be used:
Straightforward comparisons between computed and experimental average structures and atomic displacement parameters can be used to identify problems in biomolecular force fields. The high accuracy and precision crystallography (compared to NMR or other solution methods) makes such comparisons uniquely informative, and the statistics of such comparisons benefit from having many copies of chains in equivalent environments in the simulation.
Structural fluctuations in the simulations can be used to estimate diffuse scattering intensities, which can be compared to measurements using modern detectors. Both simulations and experiments allow one to place Bragg and non-Bragg intensities on a common, "absolute", scale, showing that a surpisingly large fraction of scattered X-rays occur between the Bragg peaks, and are usually ignored.
Simulations provide a model for density fluctuations in regions of protein disorder or bulk solvent (mainly water). Such models appear to account for solvent contributions to Bragg intensities in ways that are a systematic improvement over the procedures used in most protein structure refinement protocols.
Simulations provide a plausible, if imperfect, model for conformational heterogeneity in biomolecular crystals. We find that thermally-activated phonon-like modes, "trapped" lattice distortions, and more localized protein and solvent conformational disorder all contribute to diffuse scatter. We are still exploring ways to make the best comparisons with experimental data.
Markus J. Buehler is the McAfee Professor of Engineering at MIT, a member of the Center for Materials Science and Engineering, and the Center for Computational Science and Engineering at the Schwarzman College of Computing. Buehler pursues modeling, design and manufacturing approaches for biomaterials that offer greater resilience and controllable properties.
Title of Talk: Bioinspired Artificial Intelligence and Protein Materials by Design
Abstract: Nature produces a variety of materials with many functions, often out of simple and abundant materials, and at low energy. Such systems - examples of which include silk, tendon, bone, nacre or diatoms - provide broad inspiration for engineering. Here we explore the translation of biological composites to engineering applications, using a variety of tools including molecular modeling, AI and machine learning, and experimental synthesis and characterization. We review a series of studies focused on the mechanical behavior of materials, especially deformation and fracture, and how these phenomena can be modeled using a combination of molecular dynamics and machine learning, to generate a novel simulated evolutionary process that offers directed adaptation of biomaterial properties. We also present case studies of protein material optimization using genetic algorithms, applied to 3D printed composites, molecular design, and a translation of protein folding to music and back. We also review a close integration of music and materials, and review our recent research on a new bio-inspired compositional technique called materiomusic.
Irene Yarovsky leads Materials Modelling and Simulation group at RMIT University, Melbourne, Australia and concurrently holds a Visiting Professor position at the Department of Materials, Imperial College London, UK. After completing her PhD at Monash University, she joined BHP Research laboratories in Melbourne, where she applied computational molecular modelling to help design advanced industrial materials. At present, she is particularly interested in studying the interface between biological systems and nanomaterials as they interact in the living organisms, the environment and novel nano-bio devices for biomedical applications.
Title of Talk: MM2000-21: optimising performance of materials for industry and biomedicine
Abstract: In collaboration with our experimental partners, we employ theoretical molecular modelling to facilitate rational design and engineering of materials with optimal performance, controlled stimuli responses and environmental robustness. In this talk, a historical perspective and examples of our multiscale models that enabled in-depth understanding of fundamental interactions within various classes of industrial and biomedical materials in realistic environments will be presented in conjunction with the methodological advancements and challenges of modelling multicomponent interfacial systems with rigour and efficiency.
Tiffany Walsh earned her PhD from U. Cambridge, Chemistry, was a Glasstone fellow in the Dept of Materials at U. Oxford, and faculty at U. Warwick, Chemistry. She is Professor of Bio/Nanotechnology, Institute for Frontier Materials, Deakin. Her research interests are modelling the interface between soft matter and solid surfaces.
Title of Talk: Guiding Peptide-driven Exfoliation and Assembly of 2D Materials using Molecular Simulation
Abstract: Peptides provide a versatile platform for the generation, organisation and activation of nanomaterials in aqueous media. However, their application and use on two dimensional (2D) nanosheet structures such as graphene, h-BN and MoS2 is hampered, due to a lack of fundamental data regarding the structure/function relationships of these bio-nano interfaces. Together with experimental characterisation, molecular simulations can provide complementary insights into the structure/functional relationships of these challenging interfaces. Here, our strategy uses bioconjugate hybrids of peptides and fatty acids to exfoliate materials into 2D nanosheets. The role of molecular simulations in revealing the molecular scale characteristics of the peptide-driven exfoliation process are discussed for graphene, particularly in the role of the fatty acids in reducing defects in the exfoliated material. Umbrella sampling simulations are also used to predict the change in free energy during different stages of the peptide-driven exfoliation process. In addition, advancements in our simulation strategy to model peptide/h-BN and peptide/MoS2 interfaces are presented. This involved development of interfacial force-fields for describing bio-interactions at h-BN and MoS2 nanosheet interfaces in aqueous media, based on first-principles calculations. Replica-exchange with solute tempering (REST) molecular dynamics (MD) simulations are used to explore the contact between the peptides and the nanosheets, to guide the design of effective bioconjugates for exfoliation and assembly. Our simulations are also used to explore construction of heterogeneous sheet stacks, starting with bio-conjugate-driven adsorption of exfoliated graphene onto h-BN surfaces in water. The outcomes of our simulations provide a strong foundation for future work to design and deploy these molecular bioconjugates in the self-assembly of 2D heterostructures.
Aijun Du is a Professor in school of Chemistry and Physics at QUT. His research focuses on developing novel nanomaterials for energy, environment, and electronics applications using computational approaches. He has over 300 publications and is 2020 Clarivate Highly Cited Researcher. He received ARC Queen-Elizabeth II and Future Fellowship awards.
Title of Talk: Computational Design and Discovery of Novel Nanomaterials for Energy and Nanoelectronics Applications
Abstract: Material properties are in-principle determined by electronic structure. The exploration of exotic physics and chemistry using first-principles approaches have demonstrated great success in the discovery and design of novel materials for energy and electronics applications. In this talk, he will present some examples from his recent research showing how theoretical predictions makes contributions to the rational design of 2D materials and guide the experimental developments. These include (i) computational screening and experimental verification of the optimal 2D MXenes for efficient hydrogen evolution reaction [ACS Catalysis 7 (2017) 494; Nature Communications 8 (2017) 13907]; (ii) computational design and experimental validation of bi-metal doped perovskite materials as the cathode of solid oxide fuel cell; (iii) the prediction of stable 2D boron hydride [Angewandte Chemie 128 (2016) 10448] as proved in the most recent experiment [Science 371 (2021) 1143]; (iv) the prediction of multiferroic 2D Janus transitional metal dichalcogenide monolayer [Nano Letters 19 (2019) 1366].
Professor Woods received his PhD. from Queen’s University and was a post-doc at the Glycobiology Institute at the University of Oxford before taking a faculty position in 1995 at the Complex Carbohydrate Research Center at the University of Georgia. His research focuses on carbohydrate-protein interactions related to human diseases, and he maintains a suite of carbohydrate-modeling tools (GLYCAM-Web, www.glycam.org).
Title of Talk: GLYCAM21: Using Hamiltonian Replica Exchange MD to Validate Pyranose Ring Populations
Abstract: Since its release 15 years ago, the GLYCAM06 force field has been continually modified and expanded and is now one of the most widely employed classical force fields for molecular dynamics (MD) simulations of carbohydrate molecules, as well as other biomolecules. Over that period, major advances in computing hardware and algorithms have occurred that have extended the accessible timescale of MD simulations from sub-nanosecond to microsecond and beyond. The enhanced conformational sampling now available has led to the observation that for some simple monosaccharides, such as xylose, the populations of ring shapes produced by GLYCAM06 disagrees with experimental NMR data, while for other monosaccharides, such as glucose, the MD simulations agree well with the NMR data.
To discover the origin of this mixed performance, we undertook a fundamental revision of the GLYCAM force field, with a focus on improving the accuracy of ring dynamic properties by revisiting the valence, torsion, and partial charge approximations. To ensure convergence of the conformational sampling in GLYCAM21 we adopted the use of Hamiltonian Replica Exchange MD. We show that 100 ns HR MD provides sampling that is equivalent to 10 us conventional MD. A comparison with NMR J-couplings for pyranose rings indicates that GLYCAM21 reproduces the NMR values to within 1 Hz for more than 90% of monosaccharides, without sacrificing the generally good performance of the GLYCAM force field in terms of glycosidic conformational preferences and intermolecular interactions.
Debra Bernhardt is an ARC Australian Laureate Fellow in the Australian Institute for Bioengineering and Nanotechnology (AIBN) and School of Chemistry and Molecular Biosciences (SCMB) at The University of Queensland. Her research program focuses on theoretical and computational approaches to develop a fundamental understanding of the behaviour of matter. She applies these approaches to a wide range of problems, particularly transport in nanoscale systems, fluctuation phenomena, design of materials, energy storage and conversion. She is a Fellow of the Royal Australian Chemical Institute and Fellow of the Australian Academy of Sciences. She was the recipient of the Association of Molecular Modellers of Australasia Medal in 2017. Debra publishes as (Debra J. Searles).
Title of Talk: Selection of Materials for Nanoelectronics and Energy Storage Systems through Theory and Molecular Simulation: Carbon Nitrides
Abstract: Enhanced computational methods and greatly increase compute power are enabling more complex materials and systems to be modelled for longer times. However, it is also important to combine the output of calculations with simple models of matter in order to optimise molecular modelling. In this talk I will illustrate this using some recent studies we have carried out on two-dimensional carbon nitride materials. Two-dimensional carbon nitride materials have many applications, including use in energy storage systems and nanoelectronics. In this talk we will discuss some of the methodologies we use to study them, and provide results that have been obtained.
Alan Mark (PhD, JCSMR, ANU) held positions in The Netherlands and Switzerland before being appointed Professor of Physical Chemistry (Groningen). He returned to Australia as a Federation Fellow in 2004. He is associated with the GROMOS and GROMACS simulation packages, methodological and force field developments plus pioneering simulations of peptide folding, membrane assembly and the mechanism of action of membrane proteins.
Title of Talk: Understanding biological micro-machines: From antimicrobial peptides to viral fusion.
Abstract: At their most basic level many proteins and peptides can be thought of as mechanical components (switches, pumps, pipes, rods and motors) which come together to form functional complexes, in essence self-organized micromachines. Whether these are transmembrane pores formed by antimicrobial peptides or large multi-component viral complexes which facilitate infection, the challenge in understanding these systems is that crucial intermediates that define their mode of action are transitory. Thus, while the mechanisms of action can be speculated upon they cannot be observed directly. The talk will focus on how atomistic molecular dynamics simulations have been used to examine the mechanism of action of increasingly complex systems spaning anti-microbial peptides, type I cytokine receptors, efflux pumps and viral fusion proteins. For example, simulations of the conformational changes within the extracellular domains of the growth hormone receptor, the prolactin receptor, erythropoietin receptor and the epidermal growth factor receptor associated with the binding (or removal) of ligand. Although multiple mechanisms of action have been proposed for these systems, simulations suggest that the fundamental processed that control motions within these systems are remarkably similar. The talk will cover how simulations have both supported and challenged proposed mechanisms, the importance of the choice of model and how such the unique insights provided by simulations (when performed appropriately) can greatly facilitate the interpretation of experimental data.
Bharath received a BA and BS from UC Berkeley in EECS and Mathematics and was valedictorian of his graduating class in mathematics. He did his PhD in computer science at Stanford University where he studied the application of deep-learning to problems in drug-discovery. At Stanford, Bharath created the deepchem.io open-source project to grow the deep drug discovery open source community, co-created the moleculenet.ai benchmark suite to facilitate development of molecular algorithms, and more. Bharath’s graduate education was supported by a Hertz Fellowship, the most selective graduate fellowship in the sciences. After his PhD, Bharath co-founded Computable a startup that built better tools for collaborative dataset management. Bharath is currently the CEO of Deep Forest Sciences, a deep tech R&D company that builds AI for deep tech applications.
Bharath is also the lead author of “TensorFlow for Deep Learning: From Linear Regression to Reinforcement Learning”, a developer’s introduction to modern machine learning, with O’Reilly Media, and the lead author of “Deep Learning for the Life Sciences”
Title of Talk: Language based Pre-training for Drug Discovery
Abstract: Pretraining has taken the NLP world by storm as ever larger language models have broken successive benchmarks. In this talk, I'll review some recent work applying pretraining to scientific challenges, and in particular will discuss the challenges of pretraining for molecular machine learning. I'll introduce our new architecture, ChemBERTa, which explores the use of BERT-style pretraining for machine learning problems inspired by drug discovery applications.
Kresten Lindorff-Larsen is a Professor of Computational Protein Biophysics at the Linderstrøm-Lang Centre for Protein Science at the University of Copenhagen. He is the director of the Lundbeck Foundation BRAINSTRUC initiative in structural biology and the Novo Nordisk Foundation PRISM (Protein Interactions and Stability in Medicine and Genomics) centre. Current research interests include developing and applying computational methods for integrative structural biology, and the integration of biophysics and genomics research.
Title of Talk: Using experiments to improve simulations of intrinsically-disordered proteins
Abstract: Intrinsically disordered proteins (IDPs) and flexible regions in multi-domain proteins display substantial conformational heterogeneity. Characterizing the conformational ensembles of these proteins in solution typically requires combining data from one or more biophysical techniques with computational modelling or simulations [1,2]. Experimental data can either be used to assess the accuracy of a computational model or to refine the computational model to get a better agreement with the experimental data.
One commonly used approach is to use the experimental data to refine conformational ensembles of IDPs in a system-specific manner [2–6]. I will instead describe an approach in which we use the experimental data to refine the force field used the simulations [7,8]. I will describe a Bayesian formalism we have developed and applied to optimize and parameterize force fields by targeting experimental observables. We have used this method to parameterize a new coarse-grained model for IDPs by targeting data from small-angle scattering and nuclear magnetic resonance spectroscopy experiments on IDPs in solution . I will describe how this model enables us to study interactions between IDPs and their formation of higher-order structures in biomolecular condensates.
Dave holds Professorial roles in Chemistry & Physics, Medicinal Chemistry and Pharmacy at La Trobe, Monash, and Nottingham Universities. He spent 30 years at CSIRO applying computational chemistry, AI, and machine learning to design of drugs, nanomaterials, and biomaterials. He’s published 250 journal articles and book chapters, filed 25 patents, and received the RACI Adrien Albert and ACS Herman Skolnik awards.
Title of Talk: Application of active learning and meta learning to design of 2D materials and photocatalysts
Abstract: Machine learning methods have created a paradigm shift in the way bioactive molecules and materials are discovered and designed. With increasing challenges around global warming, environmental pollution, food security, and energy, Ai and machine learning can work hand in hand with automation, big data, and informatics to provide computational solutions for these challenges. Electrocatalysts and photocatalysis provide a new ways to split water and reduce carbon dioxide to generate clean fuels and also to remediate sources of environmental pollution. The discovery of 2D materials has opened a a new world of materials with unprecedented properties some, such as superconductivity, superlubricant properties, and highly tuneable band gaps, extending the boundaries of energy research. This paper will summarise the modelling and design of photocatalysts for water splitting using a stacking algorithm, a type of consensus meta modelling approach with excellent performance. It will also discuss the use of active learning, a type of adaptive experimental design, to the modelling and prediction of band gaps in 2D heterostructures, with potential applications in photovoltaics.
Dr Suarez-Martinez is a Senior Lecturer at the Physics Department in Curtin University. Originally from Spain, Irene completed her PhD from University of Sussex (UK) on graphite irradiation. After a post-doc at the CNRS (France), she moved to Australia in 2009 where she has secured two ARC fellowships. Her expertise includes atomistic models of carbon materials.
Title of Talk: The challenge of generating atomistic models for activated carbon
Abstract: Activated carbons are man-made disordered nanoporous materials synthesized from a carbonaceous precursor, such as wood, coal and sugars, that is further process with CO2 or steam. They are a product of high value with industrial, medical and environmental applications. Due to the nanoporosity, activated carbons are used for their strong adsorption properties in gas separation and filtration, and they are currently being explored as a potential material for hydrogen storage.
Models of nanoporous carbons, the precursor of activated carbon, are not trivial to generate. Their structure has been much discussed in the carbon community, but it is broadly agreed that they are amorphous materials with short-range order and densities of about 1.5g/cc. Their structure is locally based on defected graphene-like platelets that connect in three dimensions producing pores. Due to this nature, pores in activated carbon are often model as the space between two layers of graphene. In more advance models, oxygen is included in flat oxygenated polyaromatic molecules. Still current models are either porous carbon models (without oxygen) or oxygenated platelets (without the porosity).
Adsorption on activated carbon can only be accurate model with a realistic model that features both the curvature and the presence of oxygen. In this talk, I will review past models from simple geometrical models to advance molecular dynamics simulations. I will propose a method to generate atomistic structures of activated carbon that contain both a 3D porous structure and oxygen functionalization.