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
SN - 0939-6411
VL - 141
SP - 81
EP - 89
JO - European Journal of Pharmaceutics and Biopharmaceutics
JF - European Journal of Pharmaceutics and Biopharmaceutics
ER -