Skip to content
DraconiticResearch
Search Sources API
203,234 studies31 sourcesUpdated 37 min ago
Peer-reviewedCC BYAdvanced Intelligent Discovery

Embedded Morgan Fingerprints for More Efficient Molecular Property Predictions With Machine Learning

  • Emilio Nuñez-Andrade1,2
  • Isaac Vidal1,3,2,4
  • Rafael Gómez‐Bombarelli5,6
  • James William Ryan7,1,2
  • Francisco J. Martín‐Martínez8,1,9,2
9 affiliations
  1. 1Department of Chemistry Swansea University Swansea UK
  2. 2Swansea University
  3. 3Grupo de Modelización y Diseño Molecular Departamento de Química Orgánica Facultad de Ciencias Universidad de Granada Granada Spain
  4. 4Universidad de Granada
  5. 5Department of Materials Science and Engineering Massachusetts Institute of Technology Cambridge Massachusetts USA
  6. 6Massachusetts Institute of Technology
  7. 7Centre for Integrative Semiconductor Materials (CISM) Swansea University Swansea UK
  8. 8Department of Chemistry Faculty of Natural, Mathematical and Engineering Sciences King’s College London London UK
  9. 9King's College London
Open at sourcePDFDOICite

Abstract

High-dimensional data arrays in molecular representations pose some significant challenges for Machine Learning applications, including overfitting and computational inefficiency during training. This is particularly relevant for Morgan Fingerprint (MFP), widely used for tasks such as classification and property prediction, when dealing with large and complex molecular datasets. This article introduces embedded Morgan Fingerprint (eMFP), a method for dimensionality reduction of MFP, while preserving the key structural information of the encoded molecule. The implementation of eMFP offers an improved data representation that mitigates the risk of overfitting while enhancing model performance. Our results demonstrate that eMFP outperforms standard MFP in regression models, including Random Forest (RF), Multi-layer Perceptron (MLP), K-Neighbors Regressor (KNR), Gradient Booster Regressor (GBR) and a Deep Neural Network (DNN), across three different databases (RedDB, NFA, and QM9), with optimal compression sizes of q = 16, q = 32 and q = 64. These findings highlight the potential of eMFP as a faster and often superior alternative to MFP, offering more efficient hyperparameter optimization in regression models for molecular property predictions, and improved performance with large datasets where high-dimensional categorical data is encoded.

Versions

Related

  1. Physics-based molecular fingerprints from spectral graph theory provide efficient geometry-aware measures of chemical similarity.
    Jacob W Toney, Ayleen Y Farnood, Samir Darouich +1
    Chemical sciencePeer-reviewedCC BY-NCFull textOrganic chemistryEurope PMC
  2. TAME: Element-wise Mixture-of-Experts Fusion for Data-Efficient and Interpretable Molecular Property Prediction
    Robert P. Schiller, Christoph Weisser, Klaus-Robert Müller +2
    ChemRxivPreprintCC BYFull textGeneral chemistryChemRxiv
  3. Subtle Energy Corrections Reshape Condensed-Phase Property Predictions from a Machine-Learning Interatomic Potential
    Haihui Wang, Xinyan Wang, Jiashu Liang +5
    arXivPreprintCC BY-NC-ND 4.0Full textPhysical chemistryarXiv
  4. CoP/Co ₂ P heterostructures embedded in biomass-derived carbons as efficient electrocatalysts for hydrogen and oxygen evolution reactions
    Abdourahmane Dieng, Ahmed Subrati, Zahra Hagheh-Kavousi +5
    RSC AdvancesPeer-reviewedCC BY 4.0Full textElectrochemistry & energyHAL
  5. Crystal structure, Hirshfeld surface analysis, computational study, molecular docking predictions and biological evaluation of a new Schiff base (E)-2-((2,4-dichlorobenzylidene)amino)aniline
    Raounek Rouag, Amani Hind Benahsene, Mohamed Abdesselem Dems +4
    Journal of Molecular StructurePeer-reviewedLicence not stated2 citedOrganic chemistryHAL

Cite

BibTeXRIS
BibTeX
@article{nunezandrade2026embedded,
  title = {{Embedded Morgan Fingerprints for More Efficient Molecular Property Predictions With Machine Learning}},
  author = {Emilio Nuñez-Andrade and Isaac Vidal and Rafael Gómez‐Bombarelli and James William Ryan and Francisco J. Martín‐Martínez},
  journal = {Advanced Intelligent Discovery},
  year = {2026},
  month = oct,
  doi = {10.1002/aidi.202500134},
  url = {https://doi.org/10.1002/aidi.202500134},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - Embedded Morgan Fingerprints for More Efficient Molecular Property Predictions With Machine Learning
AU  - Emilio Nuñez-Andrade
AU  - Isaac Vidal
AU  - Rafael Gómez‐Bombarelli
AU  - James William Ryan
AU  - Francisco J. Martín‐Martínez
PY  - 2026
DA  - 2026/10/15
JO  - Advanced Intelligent Discovery
DO  - 10.1002/aidi.202500134
UR  - https://doi.org/10.1002/aidi.202500134
AB  - High-dimensional data arrays in molecular representations pose some significant challenges for Machine Learning applications, including overfitting and computational inefficiency during training. This is particularly relevant for Morgan Fingerprint (MFP), widely used for tasks such as classification and property prediction, when dealing with large and complex molecular datasets. This article introduces embedded Morgan Fingerprint (eMFP), a method for dimensionality reduction of MFP, while preserving the key structural information of the encoded molecule. The implementation of eMFP offers an improved data representation that mitigates the risk of overfitting while enhancing model performance. Our results demonstrate that eMFP outperforms standard MFP in regression models, including Random Forest (RF), Multi-layer Perceptron (MLP), K-Neighbors Regressor (KNR), Gradient Booster Regressor (GBR) and a Deep Neural Network (DNN), across three different databases (RedDB, NFA, and QM9), with optimal compression sizes of q = 16, q = 32 and q = 64. These findings highlight the potential of eMFP as a faster and often superior alternative to MFP, offering more efficient hyperparameter optimization in regression models for molecular property predictions, and improved performance with large datasets where high-dimensional categorical data is encoded.
SN  - 2943-9981
ER  -