Heatmap small p-values

Asked May 06 '15 at 13:00

Active May 06 '15 at 14:20

Viewed 1,533 times

I'm experiencing some difficulties trying to manage too small p-values derived from a Fisher test. In particular I have p-values like: 2.220000e-121, 1.260000e-26, 1.260000e-334, and so on. The problem is that when I try to generate a heatmap of the values it seems not to take into account the differences between such too small numerical values. Can anyone help me please?

Thank you in advance.

edited May 06 '15 at 14:20

whuber

322,774

asked May 06 '15 at 13:00

NewUsr_stat

2

For the heatmap, have you considered transforming the values? log(p) maybe? – Eric Farng May 06 '15 at 13:23
Yes, I tried but they become <0. – NewUsr_stat May 06 '15 at 13:30
2

If your heatmap library doesn't support negative numbers, what about abs(log(p))? – Eric Farng May 06 '15 at 13:35
Ok I'm trying. I was not sure that was statistically correct. Thank you very much!!!! – NewUsr_stat May 06 '15 at 13:54
4

It seems from your question that you are trying to compare a set of p-values. This approach might not be very useful, especially given the extremely low p-values in your examples. See this discussion before you go too far down this road. – EdM May 06 '15 at 14:51

Heatmap small p-values

0 Answers0