Machine Learning at the Interface of Polymer Science and Biology: How Far Can We Go?

Abstract

This Perspective outlines recent progress and future directions for using machine learning (ML), a data-driven method, to address critical questions in the design, synthesis, processing, and characterization of biomacromolecules. The achievement of these tasks requires the navigation of vast and complex chemical and biological spaces, difficult to accomplish with reasonable speed. Using modern algorithms and supercomputers, quantum physics methods are able to examine systems containing a few hundred interacting species and determine the probability of finding them in a particular region of phase space, thereby anticipating their properties. Likewise, modern approaches in chemistry and biomolecular simulation, supported by high performance computing, have culminated in producing data sets of escalating size and intrinsically high complexity. Hence, using ML to extract relevant information from these fields is of paramount importance to advance our understanding of chemical and biomolecular systems. At the heart of ML approaches lie statistical algorithms, which by evaluating a portion of a given data set, identify, learn, and manipulate the underlying rules that govern the whole data set. The assembly of a quality model to represent the data followed by the predictions and elimination of error sources are the key steps in ML. In addition to a growing infrastructure of ML tools to address complex problems, an increasing number of aspects related to our understanding of the fundamental properties of biomacromolecules are exposed to ML. These fields, including those residing at the interface of polymer science and biology (i.e., structure determination, de novo design, folding, and dynamics), strive to adopt and take advantage of the transformative power offered by approaches in the ML domain, which clearly has the potential of accelerating research in the field of biomacromolecules.

Integrating Machine Learning in the Coarse-Grained Molecular Simulation of Polymers

“…The application of ML techniques is having a great impact in this field, ,, for materials properties prediction and for the implementation of inverse design strategies − based on correlations extracted from experimental data and/or modeling predictions. ML applications specifically in Polymer Informatics have been recently reviewed. ,− …”

“…We wish to contextualize our contribution with respect to previous reviews that concern “conventional” coarse grained simulations of polymers, ,, without the invocation of ML techniques, or the use of machine learning in polymer modeling that involves either the use of AI in processing the results of atomistic molecular simulations or in polymer informatics, ,,, some of which will be briefly recalled in the next section (Section ). The main focus of the present perspective, lies in the integration of ML as part of molecular simulation methods, specifically on how ML has been or could be directly incorporated into the development of GC simulation schemes that can be applied for the study of bulk polymeric or polymer-based systems (Section ), rather than the two techniques being used separately in sequence, highlighting which application areas still remain unexplored and the issues that arise in this particular field of study (Section ).…”

Controlled Bioactive Delivery Using Degradable Electroactive Polymers

“…In silico methods are being developed to facilitate polymer design, synthesis, processing, and characterization. 70 Indeed, in silico toxicity screening methods have been developed to aid the design of bioactive molecules and minimize preclinical testing in vivo in animal models. We conducted in silico toxicity screening of the polymers and constituent building blocks (Table S2) using commercially available software, Derek Nexus, which identifies structural alerts for several endpoints; 71 Sarah Nexus, a statistical-based model focused on mutagenicity only; 72 and Zeneth, an expert, knowledge-based software that delivers accurate forced degradation predictions (Derek Nexus: v. 6.0.1, Nexus: 2.2.2; Sarah Nexus: v. 3.0.0, Sarah Model: 2.0; Zeneth: v. 8.1.1) that we have previously employed during the development of biomaterials.…”

LGB-Stack: Stacked Generalization with LightGBM for Highly Accurate Predictions of Polymer Bandgap

“… 22 As a result, machine learning has been applied to the design and prediction of the structure of many polymers and their properties. 23 As an example of polymer design using machine learning, Wu et al (2019) trained a molecular design algorithm that can recognize the relationship between thermal conductivity and other target properties to identify thousands of hypothetical polymers, out of which three were comparable to those of the state-of-the-art polymers in non-composite thermoplastics. 19 As an example of prediction of polymer properties, an online platform named Polymer Genome was developed by the Ramprasad group, which hosts their own machine learning models for rapid and accurate predictions of polymer properties.…”

“…QSPR is essentially a mathematical model that connects experimental property values with a set of features derived from the molecular structures . As a result, machine learning has been applied to the design and prediction of the structure of many polymers and their properties . As an example of polymer design using machine learning, Wu et al (2019) trained a molecular design algorithm that can recognize the relationship between thermal conductivity and other target properties to identify thousands of hypothetical polymers, out of which three were comparable to those of the state-of-the-art polymers in non-composite thermoplastics .…”