Molecular structural dataset of lignin macromolecule elucidating experimental structural compositions

Sudha Cheranma Devi Eswaran, Senthil Subramaniam, Udishnu Sanyal, Robert Ralló, Xiao ZhangScientific 2022 · DOI

TL;DR

This paper introduces the Lignin Structural (LGS) Dataset, a collection of 60,000 computer-generated molecular structures of milled wood lignin. This dataset aims to bridge the gap in structural data for lignin, a promising biopolymer, by providing a resource for computational modeling and predictive studies to unlock its industrial potential.

What Problem Does This Paper Address?

Lignin, a major component of plant cell walls, is a highly abundant and renewable biopolymer with significant potential for conversion into valuable chemicals and materials. However, its complex and irregular molecular structure has historically been a major hurdle for its industrial utilization. Unlike well-defined biopolymers like DNA, lignin's monomer sequence and linkage patterns are not precisely known, making it difficult to predict its properties and design targeted applications. Existing experimental characterization provides statistical insights but lacks the detailed molecular-level data needed for advanced computational modeling. This paper addresses this critical data gap by generating a comprehensive dataset of lignin structures that accurately reflect experimental observations.

Methods

The core of this work is a novel computational approach for generating lignin structures. This method employs a "sequence-controlled polymer generation" strategy. It starts by defining the primary monomer units (coniferyl and syringyl alcohols) and then uses experimentally derived connectivity rules to build polymer chains. The process involves:

  • Monomer Sequencing: Generating plausible sequences of monomers based on S/G ratios (syringyl to guaiacyl units) and degree of polymerization (DP). A recursive permutation algorithm is used to explore various arrangements.
  • Linkage Generation: Applying rules for forming interunit linkages (e.g., β-O-4, β-β, β-5) based on experimental data, considering both linear and branched structures.
  • Graph Representation: Translating the generated sequences and linkages into topological matrices (adjacency and connectivity) and directed graphs.
  • Molecular Structure Generation: Converting these graphs into canonical SMILES strings using the Chemistry Development Toolkit (CDK).
  • Validation: Crucially, the generated structures are validated against experimental data for milled wood lignin (MWL) from spruce and birch, ensuring that properties like molecular weight, bond proportions, and functional group counts match experimental findings with high accuracy (~90%).

Datasets

The primary dataset introduced in this paper is the Lignin Structural (LGS) Dataset.

  • Content: It comprises 60,000 computer-generated molecular structures of milled wood lignin (MWL).
  • Composition: The dataset includes structures for both softwood (G-type) and hardwood (SG-type) lignins.
  • Spruce MWL (G-type): 6.3K structures.
  • Birch MWL (SG-type): 53.7K structures.
  • Features: Structures cover a degree of polymerization (DP) ranging from 3 to 25 monomer units.
  • Availability: The dataset is publicly available on Figshare.
  • Comparison: The paper also references and compares its generated structures with those from the Lignin-KMC model, another computational approach for lignin structure generation.

Key Insights and Ideas

  • Bridging the Data Gap: The most significant insight is the creation of a large-scale, experimentally validated dataset of lignin molecular structures. This fills a critical void, enabling more sophisticated computational studies.
  • Sequence Control Matters: The development of a "sequence-controlled" generation approach highlights that lignin's structure isn't entirely random. By incorporating experimental constraints on monomer and linkage frequencies, the model can generate more realistic and representative structures.
  • Structural Heterogeneity is Key: The dataset captures the inherent variability in lignin, showing how branching patterns and linkage frequencies change with the degree of polymerization and wood type (softwood vs. hardwood). This heterogeneity is crucial for understanding lignin's diverse properties.
  • Computational Power for Materials Science: This work demonstrates how advanced computational methods can be leveraged to create valuable resources for materials science, accelerating research into renewable polymers.

Key Results

  • 60,000 Lignin Structures: The LGS dataset contains 60,000 unique, computer-generated lignin structures.
  • High Accuracy Match: The generated structures match experimentally determined structural compositions with approximately 90% accuracy.
  • DP Range: Structures span a degree of polymerization (DP) from 3 to 25.
  • Linkage Dominance: The β-O-4 linkage is confirmed as the predominant linkage in both softwood and hardwood lignins, consistent with experimental findings.
  • Branching Trends: The study reveals distinct branching trends: G-type structures show an abrupt increase in branching above DP 10, while SG-type structures exhibit more gradual branching.
  • Correlation Analysis: Heatmaps show strong correlations between DP and β-O-4 linkages in both G and SG types. For G-type, β-β bonds correlate with 5-5 and DBDO linkages, while for SG-type, 4-O-5 linkages show a higher correlation with DP.

Limitations and Future Work

  • Focus on MWL: The current dataset focuses on milled wood lignin (MWL). While a standard reference, it doesn't encompass the full spectrum of lignin types found in different biomass sources or processing methods.
  • Spatial Arrangement: The generated structures are primarily 2D representations. The paper notes that spatial (3D) arrangements of lignin macromolecules were not explored, which is a significant aspect for understanding physical properties.
  • Functional Group Nuances: While functional group counts are considered, the paper mentions that differences in these counts between generated and experimental structures exist, particularly for aliphatic hydroxyl groups.
  • Future Work: The authors suggest extending the model to incorporate more complex lignin types, explore spatial arrangements, and further refine functional group representation. The dataset and generation tools are intended to be updated and extended. The dataset can also serve as a foundation for simulating lignin depolymerization processes.

Summary updated

Abstract

Lignin is one of the most abundant biopolymers in nature and has great potential to be transformed into high-value chemicals. However, the limited availability of molecular structure data hinders its potential industrial applications. Herein, we present the Lignin Structural (LGS) Dataset that includes the molecular structure of milled wood lignin focusing on two major monomeric units (coniferyl and syringyl), and the six most common interunit linkages (phenylpropane β-aryl ether, resinol, phenylcoumaran, biphenyl, dibenzodioxocin, and diaryl ether). The dataset constitutes a unique resource that covers a part of lignin's chemical space characterized by polymer chains with lengths in the range of 3 to 25 monomer units. Structural data were generated using a sequence-controlled polymer generation approach that was calibrated to match experimental lignin properties. The LGS dataset includes 60 K newly generated lignin structures that match with high accuracy (~90%) the experimentally determined structural compositions available in the literature. The LGS dataset is a valuable resource to advance lignin chemistry research, including computational simulation approaches and predictive modelling.
BibTeX
@inproceedings{eswaran2022molecular,
  author = {Sudha Cheranma Devi Eswaran and Senthil Subramaniam and Udishnu Sanyal and Robert Ralló and Xiao Zhang},
  title = {{Molecular structural dataset of lignin macromolecule elucidating experimental structural compositions}},
  booktitle = {Scientific},
  year = {2022},
  doi = {10.1038/s41597-022-01709-4},
}
Figure