Chemoresistance is a known problem for the effectiveness of chemotherapy. The mechanisms that reduce drug potency are not fully understood and it can be very difficult to predict when chemoresistance may occur. Recently, researchers have been looking to the expression of the genome as a predictor of the response of a patient to a certain drug. Pharmacogenomics could lead to an individually designed chemotherapy regime for each patient.
Paola Lecca at The Microsoft Research–University of Trento Centre for Computational and Systems Biology integrates in three ways in this method for inferring potential chemoresistance from genome expression. The work aims to address the correlations found between genes and enzymes that determine resistance and sensitivity to a drug. This method:
– Integrates known network information into the model;
– Integrates simulation of both the gene and metabolic networks into the inference network and;
– Integrates all of the inferred networks into a large network to show the correlations between genes and metabolism.
The method used here is based on Bayesian inference, which is most helpful with a large data set and enables the inclusion of known information to give models to explain the data. The method works by assigning probabilities to the hypotheses. This is alternated with simulation of inferred networks. This is then inputted into a correlation-based inference proceeding to integrate the networks.
Lecca concentrates on gemcitabine, a pancreatic cancer drug, to demonstrate and validate this new method. It is used to infer correlations between four genes responsible for resistance and the enzymes metabolising gemcitabine in pancreatic cells.
Read this fascinating and complex work in detail now, as this article is free to access for the next 4 weeks*:
An integrative network inference approach to predict mechanisms of cancer chemoresistance
*Free access to individuals is provided through an RSC Publishing personal account. Registration is quick, free and simple