|
Abstract:
|
In the conventional statistical data analysis, to keep the model parsimonious, most of the correlated variables are usually eliminated except for one. However, in genomics RNA sequence data analysis, this is not appropriate because the genes are usually working together in a modular consists of several highly-correlated genes. To eliminate most of the genes except for one does not fully reflect the natural of functionality of genomics data. In this research, we demonstrate how this issue can be dealt with using network analysis, which groups the correlated genes into a modular and use that to evaluate the effect on the outcome variables. Similar approaches can be applied to drug safety data as the adverse events are usually correlated. Using network analysis to analyze drug safety data provides a more in-depth examination about how the treatment affects drug safety. Data from clinical trials will be used to illustrate the approaches.
|
