Unsupervised Learning Unravels the Structure of Four-Arm and Linear Block Copolymer Micelles

Therapeutic Peptides Are Preferentially Solubilized in Specific Microenvironments within PEG-PLGA Polymer Nanoparticles

Polymeric nanoparticles are a highly promising drug delivery formulation. However, a lack of understanding of the molecular mechanisms that underlie their drug solubilization and controlled release capabilities has hindered the efficient clinical translation of such technologies. Polyethylene glycol-poly(lactic-co-glycolic) acid (PEG-PLGA) nanoparticles have been widely studied as cancer drug delivery vehicles. In this letter, we use unbiased coarse-grained molecular dynamics simulations to model the self-assembly of a PEG-PLGA nanoparticle and its solubulization of the anticancer peptide, EEK, with good agreement with previously reported experimental structural data. We applied unsupervised machine learning techniques to quantify the conformations that polymers adopt at various locations within the nanoparticle. We find that the local microenvironments formed by the various polymer conformations promote preferential EEK solubilization within specific regions of the NP. This demonstrates that these microenvironments are key in controlling drug storage locations within nanoparticles, supporting the rational design of nanoparticles for therapeutic applications.

Topology-controlled self-assembly of amphiphilic block copolymers

Contemporary synthetic chemistry approaches can be used to yield a range of distinct polymer topologies with precise control. The topology of a polymer strongly influences its self-assembly into complex nanostructures however a clear mechanistic understanding of the relationship between polymer topology and self-assembly has not yet been developed. In this work, we use atomistic molecular dynamics simulations to provide a nanoscale picture of the self-assembly of three poly(ethylene oxide)-poly(methyl acrylate) block copolymers with different topologies into micelles. We find that the topology affects the ability of the micelle to form a compact hydrophobic core, which directly affects its stability. Also, we apply unsupervised machine learning techniques to show that the topology of a polymer affects its ability to take a conformation in response to the local environment within the micelles. This work provides foundations for the rational design of polymer nanostructures based on their underlying topology.

Elucidating lipid conformations in the ripple phase: Machine learning reveals four lipid populations

A new mixed radial-angular, three-particle correlation function method in combination with unsupervised machine learning was applied to examine the emergence of the ripple phase in dipalmitoylphosphatidylcholine (DPPC) lipid bilayers using data from atomistic molecular dynamics simulations of system sizes ranging from 128 to 4096 lipids. Based on the acyl tail conformations, the analysis revealed the presence of four distinct conformational populations of lipids in the ripple phases of the DPPC lipid bilayers. The expected gel-like (ordered; Lo) and fluid-like (disordered; Ld) lipids are found along with their splayed tail equivalents (Lo,s and Ld,s). These lipids differ, based on their gauche distribution and tail packing. The disordered (Ld) and disordered-splayed (Ld,s) lipids spatially cluster in the ripple in the groove side, that is, in an asymmetric manner across the bilayer leaflets. The ripple phase does not contain large numbers of Ld lipids; instead they only exist on the interface of the groove side of the undulation. The bulk of the groove side is a complex coexistence of Lo,Lo,s, and Ld,s lipids. The convex side of the undulation contains predominantly Lo lipids. Thus, the structure of the ripple phase is neither a simple coexistence of ordered and disordered lipids nor a coexistence of ordered interdigitating gel-like (Lo) and ordered-splayed (Lo,s) lipids, but instead a coexistence of an ordered phase and a complex mixed phase. Principal component analysis further confirmed the existence of the four lipid groups.

Conformational Heterogeneity and Interchain Percolation Revealed in an Amorphous Conjugated Polymer

Conjugated polymers are employed in a variety of application areas due to their bright fluorescence and strong biocompatibility. However, understanding the structure of amorphous conjugated polymers on the nanoscale is extremely challenging compared to their related crystalline phases. Using a bespoke classical force field, we study amorphous poly(9,9-di-n-octylfluorene-alt-benzothiadiazole) (F8BT) with molecular dynamics simulations to investigate the role that its nanoscale structure plays in controlling its emergent (and all-important) optical properties. Notably, we show that a giant percolating cluster exists within amorphous F8BT, which has ramifications in understanding the nature of interchain species that drive the quantum yield reduction and bathochromic shift observed in conjugated polymer-based devices and nanostructures. We also show that distinct conformations can be unravelled from within the disordered structure of amorphous F8BT using a two-stage machine learning protocol, highlighting a link between molecular conformation and ring stacking propensity. This work provides predictive understanding by which to enhance the optical properties of next-generation conjugated polymer-based devices and materials by rational, simulation-led design principles.

Predicting aggregate morphology of sequence-defined macromolecules with recurrent neural networks

Self-assembly of dilute sequence-defined macromolecules is a complex phenomenon in which the local arrangement of chemical moieties can lead to the formation of long-range structure. The dependence of this structure on the sequence necessarily implies that a mapping between the two exists, yet it has been difficult to model so far. Predicting the aggregation behavior of these macromolecules is challenging due to the lack of effective order parameters, a vast design space, inherent variability, and high computational costs associated with currently available simulation techniques. Here, we accurately predict the morphology of aggregates self-assembled from sequence-defined macromolecules using supervised machine learning. We find that regression models with implicit representation learning perform significantly better than those based on engineered features such as k-mer counting, and a recurrent-neural-network-based regressor performs the best out of nine model architectures we tested. Furthermore, we demonstrate the high-throughput screening of monomer sequences using the regression model to identify candidates for self-assembly into selected morphologies. Our strategy is shown to successfully identify multiple suitable sequences in every test we performed, so we hope the insights gained here can be extended to other increasingly complex design scenarios in the future, such as the design of sequences under polydispersity and at varying environmental conditions.

Machine learning approaches for biomolecular, biophysical, and biomaterials research

A fluent conversation with a virtual assistant, person-tailored news feeds, and deep-fake images created within seconds-all those things that have been unthinkable for a long time are now a part of our everyday lives. What these examples have in common is that they are realized by different means of machine learning (ML), a technology that has fundamentally changed many aspects of the modern world. The possibility to process enormous amount of data in multi-hierarchical, digital constructs has paved the way not only for creating intelligent systems but also for obtaining surprising new insight into many scientific problems. However, in the different areas of biosciences, which typically rely heavily on the collection of time-consuming experimental data, applying ML methods is a bit more challenging: Here, difficulties can arise from small datasets and the inherent, broad variability, and complexity associated with studying biological objects and phenomena. In this Review, we give an overview of commonly used ML algorithms (which are often referred to as "machines") and learning strategies as well as their applications in different bio-disciplines such as molecular biology, drug development, biophysics, and biomaterials science. We highlight how selected research questions from those fields were successfully translated into machine readable formats, discuss typical problems that can arise in this context, and provide an overview of how to resolve those encountered difficulties.

Impact of drug aggregation on the structural and dynamic properties of Triton X-100 micelles

Surfactants are used in a wide range of chemical and biological applications, and for pharmaceutical purposes are frequently employed to enhance the solubility of poorly water soluble drugs. In this study, all-atom molecular dynamics (MD) simulations and small-angle neutron scattering (SANS) experiments have been used to investigate the drug solubilisation capabilities of the micelles that result from 10 wt% aqueous solutions of the non-ionic surfactant, Triton X-100 (TX-100). Specifically, we have investigated the solubilisation of saturation amounts of the sodium salts of two nonsteroidal anti-inflammatory drugs: ibuprofen and indomethacin. We find that the ibuprofen-loaded micelles are more non-spherical than the indomethacin-loaded micelles which are in turn even more non-spherical than the TX-100 micelles that form in the absence of any drug. Our simulations show that the TX-100 micelles are able to solubilise twice as many indomethacin molecules as ibuprofen molecules, and the indomethacin molecules form larger aggregates in the core of the micelle than ibuprofen. These large indomethacin aggregates result in the destabilisation of the TX-100 micelle, which leads to an increase in the amount of water inside of the core of the micelle. These combined effects cause the eventual division of the indomethacin-loaded micelle into two daughter micelles. These results provide a mechanistic description of how drug interactions can affect the stability of the resulting nanoparticles.

Unsupervised learning of sequence-specific aggregation behavior for a model copolymer

We apply a recently developed unsupervised machine learning scheme for local environments [Reinhart, Comput. Mater. Sci., 2021, 196, 110511] to characterize large-scale, disordered aggregates formed by sequence-defined macromolecules. This method provides new insight into the structure of these disordered, dilute aggregates, which has proven difficult to understand using collective variables manually derived from expert knowledge [Statt et al., J. Chem. Phys., 2020, 152, 075101]. In contrast to such conventional order parameters, we are able to classify the global aggregate structure directly using descriptions of the local environments. The resulting characterization provides a deeper understanding of the range of possible self-assembled structures and their relationships to each other. We also provide detailed analysis of the effects of finite system size, stochasticity, and kinetics of these aggregates based on the learned collective variables. Interestingly, we find that the spatiotemporal evolution of systems in the learned latent space is smooth and continuous, despite being derived from only a single snapshot from each of about 1000 monomer sequences. These results demonstrate the insight which can be gained by applying unsupervised machine learning to soft matter systems, especially when suitable order parameters are not known.