Improving Small Molecule pKa Prediction Using Transfer Learning With Graph Neural Networks

Fritz Mayr; Marcus Wieder; Oliver Wieder; Thierry Langer

doi:10.1101/2022.01.20.476787

Improving Small Molecule pKa Prediction Using Transfer Learning With Graph Neural Networks

Fritz Mayr
, Marcus Wieder
, Oliver Wieder
, Thierry Langer

Department of Pharmaceutical Sciences

Publications: Contribution to journal › Article › Peer Reviewed

Abstract

Enumerating protonation states and calculating microstate pK _a values of small molecules is an important yet challenging task for lead optimization and molecular modeling. Commercial and non-commercial solutions have notable limitations such as restrictive and expensive licenses, high CPU/GPU hour requirements, or the need for expert knowledge to set up and use. We present a graph neural network model that is trained on 714,906 calculated microstate pK _a predictions from molecules obtained from the ChEMBL database. The model is fine-tuned on a set of 5,994 experimental pK _a values significantly improving its performance on two challenging test sets. Combining the graph neural network model with Dimorphite-DL, an open-source program for enumerating ionization states, we have developed the open-source Python package pkasolver, which is able to generate and enumerate protonation states and calculate pK _a values with high accuracy.

Original language	English
Article number	866585
Journal	Frontiers in Chemistry
Volume	10
DOIs	https://doi.org/10.1101/2022.01.20.476787 https://doi.org/10.3389/fchem.2022.866585 https://doi.org/10.3389/fchem.2022.866585
Publication status	Published - 1 May 2022

Funding

MW acknowledges support from an FWF Erwin Schrödinger Postdoctoral Fellowship J 4245-N28. FM and TL gratefully acknowledge funding by the NeuroDeRisk project ( https://www.neuroderisk.eu ), which has received funding from the Innovative Medicines Initiative 2 Joint Undertaking (IMI2 JU, https://european-union.europa.eu/institutions-law-budget/institutions-and-bodies/institutions-and-bodies-profiles/imi-2-ju_en ) under Grant Agreement No. 821528. This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation program and the European Federation of Pharmaceutical Industries and Associations (EFPIA, https://www.efpia.eu ).

Austrian Fields of Science 2012

301207 Pharmaceutical chemistry

Keywords

Graph Neural Network (GNN)
physical properties
PKA
protonation states
transfer learning

Access to Document

Cite this

@article{2fc4c6c1ade145fc8c3c3130cec2d233,

title = "Improving Small Molecule pKa Prediction Using Transfer Learning With Graph Neural Networks",

abstract = "Enumerating protonation states and calculating microstate pK a values of small molecules is an important yet challenging task for lead optimization and molecular modeling. Commercial and non-commercial solutions have notable limitations such as restrictive and expensive licenses, high CPU/GPU hour requirements, or the need for expert knowledge to set up and use. We present a graph neural network model that is trained on 714,906 calculated microstate pK a predictions from molecules obtained from the ChEMBL database. The model is fine-tuned on a set of 5,994 experimental pK a values significantly improving its performance on two challenging test sets. Combining the graph neural network model with Dimorphite-DL, an open-source program for enumerating ionization states, we have developed the open-source Python package pkasolver, which is able to generate and enumerate protonation states and calculate pK a values with high accuracy. ",

keywords = "Graph Neural Network (GNN), physical properties, PKA, protonation states, transfer learning",

author = "Fritz Mayr and Marcus Wieder and Oliver Wieder and Thierry Langer",

year = "2022",

month = may,

day = "1",

doi = "10.1101/2022.01.20.476787",

language = "English",

volume = "10",

journal = "Frontiers in Chemistry",

issn = "2296-2646",

publisher = "FRONTIERS MEDIA SA",

}

TY - JOUR

T1 - Improving Small Molecule pKa Prediction Using Transfer Learning With Graph Neural Networks

AU - Mayr, Fritz

AU - Wieder, Marcus

AU - Wieder, Oliver

AU - Langer, Thierry

PY - 2022/5/1

Y1 - 2022/5/1

N2 - Enumerating protonation states and calculating microstate pK a values of small molecules is an important yet challenging task for lead optimization and molecular modeling. Commercial and non-commercial solutions have notable limitations such as restrictive and expensive licenses, high CPU/GPU hour requirements, or the need for expert knowledge to set up and use. We present a graph neural network model that is trained on 714,906 calculated microstate pK a predictions from molecules obtained from the ChEMBL database. The model is fine-tuned on a set of 5,994 experimental pK a values significantly improving its performance on two challenging test sets. Combining the graph neural network model with Dimorphite-DL, an open-source program for enumerating ionization states, we have developed the open-source Python package pkasolver, which is able to generate and enumerate protonation states and calculate pK a values with high accuracy.

AB - Enumerating protonation states and calculating microstate pK a values of small molecules is an important yet challenging task for lead optimization and molecular modeling. Commercial and non-commercial solutions have notable limitations such as restrictive and expensive licenses, high CPU/GPU hour requirements, or the need for expert knowledge to set up and use. We present a graph neural network model that is trained on 714,906 calculated microstate pK a predictions from molecules obtained from the ChEMBL database. The model is fine-tuned on a set of 5,994 experimental pK a values significantly improving its performance on two challenging test sets. Combining the graph neural network model with Dimorphite-DL, an open-source program for enumerating ionization states, we have developed the open-source Python package pkasolver, which is able to generate and enumerate protonation states and calculate pK a values with high accuracy.

KW - Graph Neural Network (GNN)

KW - physical properties

KW - PKA

KW - protonation states

KW - transfer learning

UR - https://www.scopus.com/pages/publications/85133227982

U2 - 10.1101/2022.01.20.476787

DO - 10.1101/2022.01.20.476787

M3 - Article

SN - 2296-2646

VL - 10

JO - Frontiers in Chemistry

JF - Frontiers in Chemistry

M1 - 866585

ER -

Improving Small Molecule pKa Prediction Using Transfer Learning With Graph Neural Networks

Abstract

Funding

Austrian Fields of Science 2012

Keywords

Access to Document

Other files and links

Fingerprint

Cite this