Gene Selection

Microarray analysis produces many of thousands of datapoints per array, the vast majority of which is non-informative. The first and most important step in data analysis is the selection of statistically significant and biologically relevant genes of interest.

All downstream analysis is based on these lists, thus we place a great deal of importance on the methods by which these are selected.

Our Approach:

We carry out extensive data quality control including PCA and hierarchical clustering.
We will filter the data based on intensity p-value, background and signature p-value based on the ratio between two conditions.
Rather than using arbitrary values, we utilise fold change filtering based on the coefficient of variation of the background.
We finally apply a multiplicity test correction p-value filter to ensure the lists provided are as statistically relevant as possible.