🤖 AI Summary
Cell reprogramming strategy discovery is time-consuming, costly, and lacks systematic computational approaches. To address this, this paper proposes a deep reinforcement learning (DRL)-based computational framework. Methodologically, it models asynchronous Boolean/probabilistic Boolean network dynamics as a controllable agent migration task across an attractor landscape—introducing, for the first time, the concept of “pseudo-attractors” and an online detection mechanism. This reformulates the high-dimensional, non-convex, discrete biological control problem into a trainable DRL optimization problem. Leveraging Deep Q-Networks (DQN) integrated with Boolean dynamics simulation and real-time pseudo-attractor identification, the framework successfully discovers efficient and robust reprogramming intervention sequences across multiple canonical cell-fate decision models. Experimental results demonstrate substantial performance gains over conventional heuristic and random search methods, validating both efficacy and generalizability.
📝 Abstract
Cellular reprogramming can be used for both the prevention and cure of different diseases. However, the efficiency of discovering reprogramming strategies with classical wet-lab experiments is hindered by lengthy time commitments and high costs. In this study, we develop a novel computational framework based on deep reinforcement learning that facilitates the identification of reprogramming strategies. For this aim, we formulate a control problem in the context of cellular reprogramming for the frameworks of BNs and PBNs under the asynchronous update mode. Furthermore, we introduce the notion of a pseudo-attractor and a procedure for identification of pseudo-attractor state during training. Finally, we devise a computational framework for solving the control problem, which we test on a number of different models.