Evidential deep learning for trustworthy prediction of enzyme commission number

Han, So-Ra; Park, Mingyu; Kosaraju, Sai; Lee, JeungMin; Lee, Hyun; Lee, Jun Hyuck; Oh, Tae-Jin; Kang, Mingon

doi:10.1093/bib/bbad401

Cited by 4 publications

(1 citation statement)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In particular, the key for machine learning-based protein structure prediction is the extraction of essential structural features from protein data, which serve as the basis for predicting the threedimensional arrangement of atoms in a protein molecule [44][45][46][47][48][49][50][51][52][53]. Take AlphaFold for example, which is an artificial intelligence system developed by DeepMind, a subsidiary of Alphabet Inc. (Google's parent company).…”

Section: Introductionmentioning

confidence: 99%

A Reversible Spherical Geometric Conversion of Protein Backbone Structure Coordinate Matrix to Three Independent Vectors of ρ, θ and φ

2024

Preprint

View full text Add to dashboard Cite

Due to the vast conformational space proteins can adopt, accurate and efficient prediction of protein structure remains still a challenging task, coupled with the intricacies of interatomic interactions and the limitations of current computational models in effectively navigating this complex molecular landscape. Additionally, the lack of comprehensive experimental data for all protein structures further exacerbates the difficulty in reliable machine learning-based prediction of the three-dimensional conformations of the proteios building block of life. Geometrically, Cartesian coordinate system (CCS, X, Y and Z) and spherical coordinate system (SCS, ρ, θ and φ) are two interconvertible coordinate systems, and are like two sides of one coin. Since the beginning of Protein Data Bank (PDB) in 1971, CCS has been the default approach to specify atomic positions with X, Y and Z in PDB. In this manuscript, therefore, I present a novel method for the reversible spherical geometric conversion of protein backbone structure coordinate matrices to three independent vectors: ρ, θ and φ. This reversible conversion facilitates lossless extraction of essential structural features from protein backbone structural data, enabling the development of advanced novel algorithms for protein structure prediction in future. In short, this inter-atomic SCS approach offers a comprehensive yet efficient means of representing protein backbone geometry, leveraging spherical coordinates to capture spatial relationships in a compact and intuitive inter-atomic manner, and to provide a robust framework for reversible feature extraction for the ongoing efforts in advancing the field of protein structure prediction, the holy grail of computational structural biology.

show abstract

Section: Introductionmentioning

confidence: 99%

A Reversible Spherical Geometric Conversion of Protein Backbone Structure Coordinate Matrix to Three Independent Vectors of ρ, θ and φ

2024

Preprint

View full text Add to dashboard Cite

show abstract

Genome Annotation: Perspective From Bacterial Genomes

Lim,

Young

2024

Reference Module in Life Sciences

View full text Add to dashboard Cite

Accurately predicting enzyme functions through geometric graph learning on ESMFold-predicted structures

Song,

Yuan,

Chen

et al. 2024

Nat Commun

View full text Add to dashboard Cite

Enzymes are crucial in numerous biological processes, with the Enzyme Commission (EC) number being a commonly used method for defining enzyme function. However, current EC number prediction technologies have not fully recognized the importance of enzyme active sites and structural characteristics. Here, we propose GraphEC, a geometric graph learning-based EC number predictor using the ESMFold-predicted structures and a pre-trained protein language model. Specifically, we first construct a model to predict the enzyme active sites, which is utilized to predict the EC number. The prediction is further improved through a label diffusion algorithm by incorporating homology information. In parallel, the optimum pH of enzymes is predicted to reflect the enzyme-catalyzed reactions. Experiments demonstrate the superior performance of our model in predicting active sites, EC numbers, and optimum pH compared to other state-of-the-art methods. Additional analysis reveals that GraphEC is capable of extracting functional information from protein structures, emphasizing the effectiveness of geometric graph learning. This technology can be used to identify unannotated enzyme functions, as well as to predict their active sites and optimum pH, with the potential to advance research in synthetic biology, genomics, and other fields.

show abstract

Evidential deep learning for trustworthy prediction of enzyme commission number

Cited by 4 publications

References 39 publications

A Reversible Spherical Geometric Conversion of Protein Backbone Structure Coordinate Matrix to Three Independent Vectors of ρ, θ and φ

A Reversible Spherical Geometric Conversion of Protein Backbone Structure Coordinate Matrix to Three Independent Vectors of ρ, θ and φ

Genome Annotation: Perspective From Bacterial Genomes

Accurately predicting enzyme functions through geometric graph learning on ESMFold-predicted structures

Contact Info

Product

Resources

About