Abstract
Carcinogenesis is one example of transition between two biological states involving strong gene expression rearrangements [1]. Normal and tumor states are understood as two different states of a Gene Regulatory Network [2]. In terms of the normal state characteristics, the transition may be thought as a deregulation process [3,4,5]. According to a very abstract model for the transition, the available data on cancer risk support the idea of a single big jump in expression space [6], which may be associated to a deregulation cascade. Here, we use the measured frequency distribution of gene deregulations, and concepts from the probabilistic theory of causation [7] in order to infer causal relations between pairs of genes and construct a deregulation causal network [8]. Then, the deregulated genes in each sample are organized according to the network in such a way that they show the deregulation cascade behind the transition from the normal to the tumor state. The explicit analysis for prostate adenocarcinoma is given. In most cases, the cascade happens to be unique and initiated by a single deregulation event. Using results from a companion paper [9], we show that cascades conform classes which may be labeled by the deregulations in a predefined panel of 15 genes. The results of the paper may be checked in experiments with cellular lines or in animal models, and could have an impact on personalized cancer therapy.
Publisher
Cold Spring Harbor Laboratory