Connectome-constrained ligand-receptor interaction analysis for understanding brain network communication

Both diffusion magnetic resonance imaging and transcriptomic technologies have provided unprecedented opportunities to dissect brain network communication, offering insights from structural connectivity and signaling molecules separately. However, incorporating these complementary modalities for characterizing the interregional communication remains challenging. By simplifying the communication processes into an optimal transport problem, which is defined as the ligand-receptor expression coupling constrained by structurally-derived communication cost, we develop a method called CLRIA (connectome-constrained ligand-receptor interaction analysis) to infer a low-rank representation of ligand-receptor interaction-mediated communication networks. To solve the proposed optimization problem, the block majorization minimization algorithm is adopted and proven to converge globally. We benchmark CLRIA on simulated and published data, validating its accuracy and computational efficiency. Subsequently, we demonstrate the CLRIA's utility in evaluating communication strategies and asymmetric communication using its solution. Furthermore, CLRIA-derived communication patterns successfully decode brain state transitions. Overall, our results highlight CLRIA as a valuable tool for understanding complex communication in the brain. By modeling the brain communication as an optimal transport problem, CLRIA integrates structural and molecular data, and reveals the bias for hybrid strategy, asymmetric hierarchical signaling, and molecular perspective for state transitions.