Interpret your ranking

In principle, the presented scores have one feature in common: they are distributed around 0. A high positive value indicates up-regulation or corre lation, a high negative value indicates down-regulation or anti-correlation. To rank for differential gene expression in general, take absolute scores which is abs(score) in R.

If you want, you can stop here. You have a ranking of all genes on the array and you can pick genes from the top of the list as you like. This is exploratory data analysis, and there is nothing wrong with it. You might say: 'I can handle 50 genes. Hence, I take the top 50 genes.' This is a practical argument, and we do not see any problems with it, too. However, you might feel some discomfort once you realize that there are always 50 genes on top of a ranked list. Even in cases with no differentially expressed genes at all. For relief, you need to follow the track to statistical analysis.

0 0

Post a comment