癌症的发展和进展受到癌症驱动基因的重大影响。了解癌症驱动基因和它们的作用机制对于开发有效的癌症治疗方案至关重要。因此，确定驱动基因对于药物开发、癌症诊断和治疗非常重要。在此，我们提出了一种基于两阶段随机游走重启（RWR）和修改的随机游走算法中转移概率矩阵计算方法的算法来发现驱动基因。首先，在整个基因相互作用网络上执行了第一阶段的RWR，我们采用了一种新的转移概率矩阵计算方法，并提取了与种子节点有高相关性的节点的子网络。然后，将子网络应用于第二阶段的RWR，并重新对子网络中的节点进行排序。我们的方法在识别驱动基因方面优于现有方法。同时比较了三个基因相互作用网络的效果、两轮随机游走和种子节点的敏感性。此外，我们鉴定了几个潜在的驱动基因，其中一些参与驱动癌症的发展。总体而言，我们的方法在各种癌症类型中都非常有效，明显优于现有的方法，并能够确定可能的驱动基因。 版权所有©2023 Elsevier Ltd.

Cancer development and progression are significantly influenced by cancer driver genes. Understanding cancer driver genes and their mechanisms of action is essential for developing effective cancer treatments. As a result, identifying driver genes is important for drug development, cancer diagnosis, and treatment. Here, we present an algorithm to discover driver genes based on the two-stage random walk with restart (RWR), and the modified method for calculating the transition probability matrix in random walk algorithm. First, we performed the first stage of RWR on the whole gene interaction network, in which we employ a new method for calculating the transition probability matrix and extracted the subnetwork based on nodes that had a high correlation with the seed nodes. The subnetwork was then applied to the second stage of RWR and the nodes were re-ranked in the subnetwork. Our approach outperformed existing methods in identifying driver genes. The outcome of the effect of three gene interaction networks, two rounds of random walk, and the seed nodes' sensitivity were all compared at the same time. In addition, we identified several potential driver genes, some of which are involved in driving cancer development. Overall, our method is efficient in various cancer types, significantly outperforms existing methods, and can identify possible driver genes.Copyright © 2023 Elsevier Ltd. All rights reserved.