Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Automatic single- and multi-label enzymatic function prediction by machine learning.

Shervine Amidi1, Afshine Amidi1, Dimitrios Vlachakis2

  • 1Department of Applied Mathematics, Center for Visual Computing, Ecole Centrale de Paris (CentraleSupélec), Châtenay-Malabry, France.

Peerj
|April 4, 2017
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Metabolic determinants of cancer immunotherapy outcomes identified by plasma profiling.

Nature medicine·2026
Same author

Novel Disease-Specific Panel of Salivary microRNAs for the Detection of Oral Squamous Cell Carcinoma from Early Invasion to Stage IV Disease.

International journal of molecular sciences·2026
Same author

Evaluation of Two Commercial Artificial Intelligence Segmentation Systems for Radiation Therapy.

Journal of medical physics·2026
Same author

DNA-PK interacts with cyclic dinucleotides and inhibits type I interferon responses.

The Journal of experimental medicine·2026
Same author

mpMRI-Based Risk Estimation to Optimize Prostate Cancer Patient Selection for Active Surveillance.

Cancers·2026
Same author

Stress-related disorders and nonpeptidic CRH antagonists.

Hormones (Athens, Greece)·2026
Same journal

Association between intestinal functional disorders and anal fistula: evidence from a retrospective case-control study.

PeerJ·2026
Same journal

Automated recognition of Meso-Cenozoic foraminifera from Senegalese sedimentary deposits using convolutional neural networks.

PeerJ·2026
Same journal

Genome-wide analysis of <i>HSP70</i> gene superfamily in kelp (<i>Saccharina japonica</i>): identification, characterization, and heat stress-responsive expression profiles.

PeerJ·2026
Same journal

Morphological and molecular evidence of the Antarctic sleeper shark <i>Somniosus antarcticus</i> (Somniosidae) in northern Chile.

PeerJ·2026
Same journal

Stroboscopic balance training enhances dynamic stability and postural control in collegiate badminton players: a randomized controlled trial.

PeerJ·2026
Same journal

Frequent exposure to biologics is associated with small intestinal bacterial overgrowth in patients with Crohn's disease: a retrospective case-control study.

PeerJ·2026
See all related articles

This study introduces a new multi-label classification model for predicting enzymatic function using protein structure and sequence data. The model accurately identifies multiple functions for enzymes, improving upon previous single-label methods.

Area of Science:

  • Biochemistry
  • Bioinformatics
  • Computational Biology

Background:

  • The Protein Data Bank (PDB) has seen a >15-fold increase in protein structures since 1999.
  • Predicting enzymatic function computationally is crucial for understanding enzyme behavior in catalyzing reactions.
  • Existing single-label classification methods are limited, especially for multi-functional enzymes.

Purpose of the Study:

  • To develop a multi-label enzymatic function classification scheme.
  • To combine structural and amino acid sequence information for improved prediction.
  • To assess the methodology for general enzymatic function prediction (EC code first digit).

Main Methods:

  • A multi-label classification scheme was developed.
  • Two fusion approaches were investigated: feature-level and decision-level.
Keywords:
Amino acid sequenceEnzyme classificationMulti-labelSingle-labelSmith-Waterman algorithmStructural information

Related Experiment Videos

  • The models were trained and tested on 40,034 enzymes from the PDB database.

    • Main Results:

    • The proposed single-label model achieved 97.8% accuracy (Hamming-loss).
    • The multi-label model achieved 95.5% accuracy (Hamming-loss).
    • The multi-label model predicted all possible enzymatic reactions for 85.4% of multi-labeled enzymes.

    Conclusions:

    • The developed multi-label classification scheme effectively predicts enzymatic functions, including multiple functions per enzyme.
    • Combining structural and sequence data offers a robust approach for enzymatic function prediction.
    • The study provides valuable computational tools and datasets for enzyme research.