Dataset Information

Machine learning strategies for the structure-property relationship of copolymers.

ABSTRACT: Establishing the structure-property relationship is extremely valuable for the molecular design of copolymers. However, machine learning (ML) models can incorporate both chemical composition and sequence distribution of monomers, and have the generalization ability to process various copolymer types (e.g., alternating, random, block, and gradient copolymers) with a unified approach are missing. To address this challenge, we formulate four different ML models for investigation, including a feedforward neural network (FFNN) model, a convolutional neural network (CNN) model, a recurrent neural network (RNN) model, and a combined FFNN/RNN (Fusion) model. We use various copolymer types to systematically validate the performance and generalizability of different models. We find that the RNN architecture that processes the monomer sequence information both forward and backward is a more suitable ML model for copolymers with better generalizability. As a supplement to polymer informatics, our proposed approach provides an efficient way for the evaluation of copolymers.

SUBMITTER: Tao L

PROVIDER: S-EPMC9249671 | biostudies-literature | 2022 Jul

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning strategies for the structure-property relationship of copolymers.

Tao Lei L Byrnes John J Varshney Vikas V Li Ying Y

iScience 20220610 7

Establishing the structure-property relationship is extremely valuable for the molecular design of copolymers. However, machine learning (ML) models can incorporate both chemical composition and sequence distribution of monomers, and have the generalization ability to process various copolymer types (e.g., alternating, random, block, and gradient copolymers) with a unified approach are missing. To address this challenge, we formulate four different ML models for investigation, including a feedfo ...[more]

PMID: 35789847

Similar Datasets

Project description:Poly(2-oxazoline)s (POZs) are widely investigated for their applications in various fields due to their unique properties. To exploit and combine different characteristics of the POZ family, 2-oxazoline monomers can be copolymerized to prepare tailor-made copolymers with the desired glass transition temperature (T g), melting temperature (T m), amphiphilicity, and functionality. Here, we report the synthesis and characterization of 2-oxazoline monomers and a range of POZ copolymers produced, thereof. 2-Propyl-2-oxazoline (PrOZ) and 2-pentyl-2-oxazoline (PeOZ) monomers were synthesized by two different methods starting from nitriles or carboxylic acids. A number of POZ copolymers were synthesized by copolymerization of 2-ethyl-2-oxazoline (EOZ) with either one of PrOZ, PeOZ, or 2-phenyl-2-oxazoline (PhOZ) at three different compositions (25:75, 50:50, and 75:25) and three molecular weights (1000, 2000, and 5000 Da). The successful synthesis of the monomers and copolymers was demonstrated through their structural analysis by 1H NMR and FTIR. SEC results confirmed the targeted molar masses of the copolymers and living nature of the polymerization by showing low dispersity values. Thermal properties of the copolymers were studied using DSC and TGA. DSC studies revealed the amorph and random state of the copolymers with obtained T g values for the copolymers in the range of -3 to 84 °C depending on their molecular weight and type of the side chain. While the presence of longer aliphatic side chains resulted in lower T g values, the presence of 2-phenyl substituents on the polymer led to higher T g values. The decomposition temperatures determined by TGA were in the range of 328 to 383 °C depending on the molecular weight, composition, and side chain of the copolymers. It was observed that higher molecular weights led to higher T g values and decomposition temperatures. While copolymers with aliphatic side chains exhibited a single-step decomposition profile, the decomposition of copolymers having aromatic side chains occurred in multiple steps. The variations in the molecular weight, composition, and side chains of the copolymers resulted in a library of tailorable amphiphilic copolymers suitable for multiple applications ranging from biomedical applications to composite manufacturing.

Project description:Waterborne polyurethane (WPU) has attracted significant interest as a promising alternative to solvent-based polyurethane (SPU) due to its positive impact on safety and sustainability. However, significant limitations of WPU, such as its weaker mechanical strength, limit its ability to replace SPU. Triblock amphiphilic diols are promising materials to enhance the performance of WPU due to their well-defined hydrophobic-hydrophilic structures. Yet, our understanding of the relationship between the hydrophobic-hydrophilic arrangements of triblock amphiphilic diols and the physical properties of WPU remains limited. In this study, we show that by controlling the micellar structure of WPU in aqueous solution via the introduction of triblock amphiphilic diols, the postcuring efficiency and the resulting mechanical strength of WPU can be significantly enhanced. Small-angle neutron scattering confirmed the microstructure and spatial distribution of hydrophilic and hydrophobic segments in the engineered WPU micelles. In addition, we show that the control of the WPU micellar structure through triblock amphiphilic diols renders WPU attractive in the applications of controlled release, such as drug delivery. Here, curcumin was used as a model hydrophobic drug, and the drug release behavior from WPU-micellar-based drug delivery systems was characterized. It was found that curcumin-loaded WPU drug delivery systems were highly biocompatible and exhibited antibacterial properties in vitro. Furthermore, the sustained release profile of the drug was found to be dependent on the structure of the triblock amphiphilic diols, suggesting the possibility of controlling the drug release profile via the selection of triblock amphiphilic diols. This work shows that by shedding light on the structure-property relationship of triblock amphiphilic diol-containing WPU micelles, we may enhance the applicability of WPU systems and move closer to realizing their promising potential in real-life applications.

Project description:In recent years machine learning has transformed many aspects of the drug discovery process including small molecule design for which the prediction of the bioactivity is an integral part. Leveraging structural information about the interactions between a small molecule and its protein target has great potential for downstream machine learning scoring approaches, but is fundamentally limited by the accuracy with which protein:ligand complex structures can be predicted in a reliable and automated fashion. With the goal of finding practical approaches to generating useful kinase:inhibitor complex geometries for downstream machine learning scoring approaches, we present a kinase-centric docking benchmark assessing the performance of different classes of docking and pose selection strategies to assess how well experimentally observed binding modes are recapitulated in a realistic cross-docking scenario. The assembled benchmark data set focuses on the well-studied protein kinase family and comprises a subset of 589 protein structures co-crystallized with 423 ATP-competitive ligands. We find that the docking methods biased by the co-crystallized ligand-utilizing shape overlap with or without maximum common substructure matching-are more successful in recovering binding poses than standard physics-based docking alone. Also, docking into multiple structures significantly increases the chance to generate a low RMSD docking pose. Docking utilizing an approach that combines all three methods (Posit) into structures with the most similar co-crystallized ligands according to shape and electrostatics proofed to be the most efficient way to reproduce binding poses achieving a success rate of 66.9 % across all included systems. The studied docking and pose selection strategies-which utilize the OpenEye Toolkit-were implemented into pipelines of the KinoML framework allowing automated and reliable protein:ligand complex generation for future downstream machine learning tasks. Although focused on protein kinases, we believe the general findings can also be transferred to other protein families.

Dataset Information

Machine learning strategies for the structure-property relationship of copolymers.

Publications

Machine learning strategies for the structure-property relationship of copolymers.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets