DeepSP: Deep learning-based spatial properties to predict monoclonal antibody stability

Abstract

Therapeutic antibody development faces challenges due to high viscosities and aggregation tendencies. The spatial charge map (SCM) and spatial aggregation propensity (SAP) are computational techniques that aid in predicting viscosity and aggregation, respectively. These methods rely on structural data derived from molecular dynamics (MD) simulations, which are computationally demanding. DeepSCM, a deep learning surrogate model based on sequence information to predict SCM, was recently developed to screen high-concentration antibody viscosity. This study further utilized a dataset of 20,530 antibody sequences to train a convolutional neural network deep learning surrogate model called Deep Spatial Properties (DeepSP). DeepSP directly predicts SAP and SCM scores in different domains of antibody variable regions based solely on their sequences without performing MD simulations. The linear correlation coefficient between DeepSP scores and MD-derived scores for 30 properties achieved values between 0.76 and 0.96 with an average of 0.87. DeepSP descriptors were employed as features to build machine learning models to predict the aggregation rate of 21 antibodies, and the performance is similar to the results obtained from the previous study using MD simulations. This result demonstrates that the DeepSP approach significantly reduces the computational time required compared to MD simulations. The DeepSP model enables the rapid generation of 30 structural properties that can also be used as features in other research to train machine learning models for predicting various antibody stability using sequences only. DeepSP is freely available as an online tool via https://deepspwebapp.onrender.com and the codes and parameters are freely available at https://github.com/Lailabcode/DeepSP.

Keywords: Antibody stability; Deep learning; Molecular dynamics simulation; Monoclonal antibody; Spatial aggregation propensity; Spatial charge map.

Graphical abstract

Fig. 1

Distribution of VH, VL, CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 lengths of the 20530 Fv sequences in this study. The CDR regions are based on the Chothia definition.

Fig. 2

Box-and-Whisker plot illustrating the normalized (rescaled to 0 −1) A) average B) standard deviation score distribution for all 30 properties obtained from MD simulations.

Fig. 3

Illustration of CNN model architecture with the training and validation loss over number of epochs for A) SAP_pos model B) SCM_neg model C) SCM_pos model, contained in DeepSP surrogate model developed in this study.

Fig. 4

Bar plot illustrating the correlation between the predicted and actual score of all 30 spatial properties.

Fig. 5

Scatter plot of correlation between predicted and experimental aggregation rate A, B) from previous study where MD simulation features were used C, D) current study where DeepSP (sequence-based) features were used.