Simulating gene silencing through intervention analysis

We propose a novel method for simulating the effects of gene silencing. Our approach combines relevant subject matter information provided by biological pathways with gene expression levels measured in regular conditions to predict the behavior of the system after one of the genes has been silenced. We achieve this by modeling gene silencing as an external intervention in a causal graphical model. To account for the uncertainty associated with the structure learning of the graphical model, we adopt a bootstrap approach. We illustrate our proposal on a Drosophila melanogaster gene silencing experiment.