Editorial: A Journey Through 50 Years of Structural Bioinformatics in Memoriam of Cyrus Chothia

Alfredo Iacoangeli, Paolo Marcatili, Charlotte M. Deane, Arthur M. Lesk, Annalisa Pastore, Sarah A. TeichmannFrontiers in Molecular Biosciences 2022 · DOI · OpenAlex

Introduction

This editorial commemorates the profound 50-year legacy of Dr. Cyrus Chothia, a foundational figure in structural bioinformatics. By curating a diverse collection of research, this Research Topic honors Chothia's pioneering work in the 1970s and 80s, which established the principles of homology modeling and protein classification. The featured studies push the boundaries of protein science, ranging from AI-driven antibody design to the identification of fundamental architectural motifs that define the known protein universe.

Addressing the Structure-Function Gap

The central challenge in structural bioinformatics is decoding the intricate relationship between a protein's primary sequence, its three-dimensional architecture, and its biological function. Despite the explosion of genomic data, predicting the dynamic behavior and evolutionary trajectories of proteins remains a monumental task. This collection addresses specific hurdles such as modeling "difficult" proteins like G protein-coupled receptors (GPCRs), understanding the nuances of antibody binding, and identifying the geometric building blocks common to all known proteins.

Methodological Innovations

The research presented reflects the state-of-the-art in computational biology, employing a variety of sophisticated techniques:

  • Molecular Dynamics (MD): Researchers utilized MD simulations to investigate nanosecond-to-millisecond transitions in antibody loops and "elbow angles," which are critical for antigen specificity.
  • Information-Theoretic Inference: This approach was applied to the Protein Data Bank (PDB) to automatically identify conserved sets of secondary structural elements.
  • Deep Learning: Advanced neural networks were used for sequence-based prediction of Solvent-Accessible Surface Area (SASA) and for mining coevolutionary information to enhance homology detection.
  • In Silico Design: Computational design was integrated with experimental methods to engineer antibodies targeting toxic protein aggregates, such as TDP-43 in Amyotrophic Lateral Sclerosis (ALS).
  • Machine Learning (ML): Specifically, "Inter-Residue Interaction Map Threading" was employed to improve the accuracy of protein homology detection.

Key Datasets and Tools

The featured research leverages cornerstone biological databases and introduces novel computational tools:

  • Protein Data Bank (PDB): The foundational source for all structural analyses.
  • ASTRAL SCOP & CATH: These hierarchical databases were essential for classifying protein domains and studying their evolutionary relationships.
  • Bio-GATS: A new, open-source tool introduced by Jabeen et al. for the homology modeling of GPCRs, available on GitHub.
  • Curated Pathogenic Variants (PVs): A specialized dataset used to map single-amino acid mutations onto protein structures to analyze solvent exposure.
  • Human Protein Sequences: A dataset of 12,494 sequences was used to validate deep learning models for surface area prediction.

Insights into the Protein Universe

One of the most striking findings in this collection is the identification of an architectural "basis set" for proteins. Researchers identified a dictionary of exactly 1,493 substructures that can describe the entire PDB, suggesting that nature utilizes a finite set of geometric "words" to construct the vast array of known proteins.

Furthermore, the research highlights that antibody geometry extends beyond the well-known CDR loops. The "elbow angle"—the orientation between variable and constant domains—is a dynamic and critical factor in target recognition. The collection also explores evolutionary constraints, noting that protein abundance correlates with residue conservation. Interestingly, disordered regions on protein surfaces show a strong evolutionary interdependency with core residues.

The editorial also touches upon the "coevolution paradox." While Cyrus Chothia famously disliked the term "coevolution," arguing that substitutions occur sequentially rather than simultaneously, the field now depends heavily on coevolutionary data for deep learning-based structure prediction.

Results and Therapeutic Implications

The contributions to this Research Topic yielded several significant results:

  • Bio-GATS Validation: The new GPCR modeling tool was validated against established benchmarks and successfully applied to a case study of an olfactory receptor.
  • SASA Prediction: Deep learning models accurately predicted solvent accessibility for over 12,000 human proteins lacking experimental 3D structures.
  • Therapeutic Potential: A study demonstrated that an in silico designed antibody could potentially sequester TDP-43 aggregates, offering a new avenue for ALS treatment.
  • Solvation Dynamics: New methods characterized the dynamic hydrogen bond network at the protein-water interface, providing insights into how proteins interact with their environment.

Future Directions

While deep learning has transformed the field, the growing volume of data presents ongoing challenges. Future research must better integrate protein sequences into structural classifications to predict functions in complex organisms. A primary frontier remains the study of protein dynamics—understanding how micro-to-millisecond movements translate into biological action. As structural bioinformatics continues to evolve, it must remain an open-minded, cross-disciplinary journey, honoring the spirit of Cyrus Chothia's career.

Summary updated

Abstract

TEST 02 - Elsevier's Scopus, the largest abstract and citation database of peer-reviewed literature. Search and access research from the science, technology, medicine, social sciences and arts and humanities fields.
BibTeX
@inproceedings{iacoangeli2022editorial,
  author = {Alfredo Iacoangeli and Paolo Marcatili and Charlotte M. Deane and Arthur M. Lesk and Annalisa Pastore and Sarah A. Teichmann},
  title = {{Editorial: A Journey Through 50 Years of Structural Bioinformatics in Memoriam of Cyrus Chothia}},
  booktitle = {Frontiers in Molecular Biosciences},
  year = {2022},
  doi = {10.3389/fmolb.2022.885318},
}
Figure