Title:Antifragility Predicts the Robustness and Evolvability of Biological Networks through Multi-class Classification with a Convolutional Neural Network

Abstract:Robustness and evolvability are essential properties to the evolution of biological networks. To determine if a biological network is robust and/or evolvable, the comparison of its functions before and after mutations is required. However, it has an increasing computational cost as network size grows. Here we aim to develop a predictor to estimate the robustness and evolvability of biological networks without an explicit comparison of functions. We measure antifragility in Boolean network models of biological systems and use this as the predictor. Antifragility is a property to improve the capability of a system through external perturbations. By means of the differences of antifragility between the original and mutated biological networks, we train a convolutional neural network (CNN) and test it to classify the properties of robustness and evolvability. We found that our CNN model successfully classified the properties. Thus, we conclude that our antifragility measure can be used as a significant predictor of the robustness and evolvability of biological networks.