Application of interpretable artificial neural networks to early monoclonal antibodies development

Lorenzo Gentiluomo*, Dierk Roessner, Dillen Augustijn, Hristo Svilenov, Alina Kulakova, Sujata Mahapatra, Gerhard Winter, Werner Streicher, Åsmund Rinnan, Günther H.J. Peters, Pernille Harris, Wolfgang Frieß

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

The development of a new protein drug typically starts with the design, expression and biophysical characterization of many different protein constructs. The initially high number of constructs is radically reduced to a few candidates that exhibit the desired biological and physicochemical properties. This process of protein expression and characterization to find the most promising molecules is both expensive and time-consuming. Consequently, many companies adopt and implement philosophies, e.g. platforms for protein expression and formulation, computational approaches, machine learning, to save resources and facilitate protein drug development. Inspired by this, we propose the use of interpretable artificial neuronal networks (ANNs) to predict biophysical properties of therapeutic monoclonal antibodies i.e. melting temperature Tm, aggregation onset temperature Tagg, interaction parameter kD as a function of pH and salt concentration from the amino acid composition. Our ANNs were trained with typical early-stage screening datasets achieving high prediction accuracy. By only using the amino acid composition, we could keep the ANNs simple which allows for high general applicability, robustness and interpretability. Finally, we propose a novel "knowledge transfer" approach, which can be readily applied due to the simple algorithm design, to understand how our ANNs come to their conclusions.
Original languageEnglish
JournalEuropean Journal of Pharmaceutics and Biopharmaceutics
Volume141
Pages (from-to)81-89
Number of pages9
ISSN0939-6411
DOIs
Publication statusPublished - 2019

Keywords

  • Machine learning
  • Monoclonal antibody
  • Neural network(s)
  • Protein aggregation
  • Protein formulation
  • Stability

Cite this

Gentiluomo, Lorenzo ; Roessner, Dierk ; Augustijn, Dillen ; Svilenov, Hristo ; Kulakova, Alina ; Mahapatra, Sujata ; Winter, Gerhard ; Streicher, Werner ; Rinnan, Åsmund ; Peters, Günther H.J. ; Harris, Pernille ; Frieß, Wolfgang. / Application of interpretable artificial neural networks to early monoclonal antibodies development. In: European Journal of Pharmaceutics and Biopharmaceutics. 2019 ; Vol. 141. pp. 81-89.
@article{81ad3f2a9fe049479120c0d085d2e2de,
title = "Application of interpretable artificial neural networks to early monoclonal antibodies development",
abstract = "The development of a new protein drug typically starts with the design, expression and biophysical characterization of many different protein constructs. The initially high number of constructs is radically reduced to a few candidates that exhibit the desired biological and physicochemical properties. This process of protein expression and characterization to find the most promising molecules is both expensive and time-consuming. Consequently, many companies adopt and implement philosophies, e.g. platforms for protein expression and formulation, computational approaches, machine learning, to save resources and facilitate protein drug development. Inspired by this, we propose the use of interpretable artificial neuronal networks (ANNs) to predict biophysical properties of therapeutic monoclonal antibodies i.e. melting temperature Tm, aggregation onset temperature Tagg, interaction parameter kD as a function of pH and salt concentration from the amino acid composition. Our ANNs were trained with typical early-stage screening datasets achieving high prediction accuracy. By only using the amino acid composition, we could keep the ANNs simple which allows for high general applicability, robustness and interpretability. Finally, we propose a novel {"}knowledge transfer{"} approach, which can be readily applied due to the simple algorithm design, to understand how our ANNs come to their conclusions.",
keywords = "Machine learning, Monoclonal antibody, Neural network(s), Protein aggregation, Protein formulation, Stability",
author = "Lorenzo Gentiluomo and Dierk Roessner and Dillen Augustijn and Hristo Svilenov and Alina Kulakova and Sujata Mahapatra and Gerhard Winter and Werner Streicher and {\AA}smund Rinnan and Peters, {G{\"u}nther H.J.} and Pernille Harris and Wolfgang Frie{\ss}",
year = "2019",
doi = "10.1016/j.ejpb.2019.05.017",
language = "English",
volume = "141",
pages = "81--89",
journal = "European Journal of Pharmaceutics and Biopharmaceutics",
issn = "0939-6411",
publisher = "Elsevier",

}

Application of interpretable artificial neural networks to early monoclonal antibodies development. / Gentiluomo, Lorenzo; Roessner, Dierk; Augustijn, Dillen; Svilenov, Hristo; Kulakova, Alina; Mahapatra, Sujata; Winter, Gerhard; Streicher, Werner; Rinnan, Åsmund; Peters, Günther H.J.; Harris, Pernille; Frieß, Wolfgang.

In: European Journal of Pharmaceutics and Biopharmaceutics, Vol. 141, 2019, p. 81-89.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Application of interpretable artificial neural networks to early monoclonal antibodies development

AU - Gentiluomo, Lorenzo

AU - Roessner, Dierk

AU - Augustijn, Dillen

AU - Svilenov, Hristo

AU - Kulakova, Alina

AU - Mahapatra, Sujata

AU - Winter, Gerhard

AU - Streicher, Werner

AU - Rinnan, Åsmund

AU - Peters, Günther H.J.

AU - Harris, Pernille

AU - Frieß, Wolfgang

PY - 2019

Y1 - 2019

N2 - The development of a new protein drug typically starts with the design, expression and biophysical characterization of many different protein constructs. The initially high number of constructs is radically reduced to a few candidates that exhibit the desired biological and physicochemical properties. This process of protein expression and characterization to find the most promising molecules is both expensive and time-consuming. Consequently, many companies adopt and implement philosophies, e.g. platforms for protein expression and formulation, computational approaches, machine learning, to save resources and facilitate protein drug development. Inspired by this, we propose the use of interpretable artificial neuronal networks (ANNs) to predict biophysical properties of therapeutic monoclonal antibodies i.e. melting temperature Tm, aggregation onset temperature Tagg, interaction parameter kD as a function of pH and salt concentration from the amino acid composition. Our ANNs were trained with typical early-stage screening datasets achieving high prediction accuracy. By only using the amino acid composition, we could keep the ANNs simple which allows for high general applicability, robustness and interpretability. Finally, we propose a novel "knowledge transfer" approach, which can be readily applied due to the simple algorithm design, to understand how our ANNs come to their conclusions.

AB - The development of a new protein drug typically starts with the design, expression and biophysical characterization of many different protein constructs. The initially high number of constructs is radically reduced to a few candidates that exhibit the desired biological and physicochemical properties. This process of protein expression and characterization to find the most promising molecules is both expensive and time-consuming. Consequently, many companies adopt and implement philosophies, e.g. platforms for protein expression and formulation, computational approaches, machine learning, to save resources and facilitate protein drug development. Inspired by this, we propose the use of interpretable artificial neuronal networks (ANNs) to predict biophysical properties of therapeutic monoclonal antibodies i.e. melting temperature Tm, aggregation onset temperature Tagg, interaction parameter kD as a function of pH and salt concentration from the amino acid composition. Our ANNs were trained with typical early-stage screening datasets achieving high prediction accuracy. By only using the amino acid composition, we could keep the ANNs simple which allows for high general applicability, robustness and interpretability. Finally, we propose a novel "knowledge transfer" approach, which can be readily applied due to the simple algorithm design, to understand how our ANNs come to their conclusions.

KW - Machine learning

KW - Monoclonal antibody

KW - Neural network(s)

KW - Protein aggregation

KW - Protein formulation

KW - Stability

U2 - 10.1016/j.ejpb.2019.05.017

DO - 10.1016/j.ejpb.2019.05.017

M3 - Journal article

C2 - 31112768

VL - 141

SP - 81

EP - 89

JO - European Journal of Pharmaceutics and Biopharmaceutics

JF - European Journal of Pharmaceutics and Biopharmaceutics

SN - 0939-6411

ER -