|
Abstract:
|
Elucidating gene regulatory network is an important step towards understanding the normal cell physiology and complex pathological phenotype. Reverse-engineering consists in using gene expression over time or over different experimental conditions to discover the structure of the gene network in a targeted cellular process. The fact that gene expression data are usually noisy, highly correlated, and have high dimensionality explains the need for specific statistical methods to reverse engineer the underlying network. Among known methods, Approximate Bayesian Computation (ABC) algorithms have not been thoroughly studied for network inference. Due to the computational overhead their application is also limited to a small number of genes. Not only have we developed a method that have less computational cost but also have we accelerated that new multi-level ABC approach by using GPUs. At the first level, the method captures the global properties of the network, such as scale-freeness and clustering coefficients, whereas the second level is targeted to capture local properties, including the probability of each couple of genes being linked. Our approach is evaluated on a real dataset.
|
