Frequently Asked Questions
How many replicate microarrays do I have to run to detect statistically significant differences?
The answer to this question is somewhat complicated. First, you must generate replicates by running duplicate microarrays or by having replicate probes on a single microarray. Included in the replicate should be one dye-swapped microarray. The minimum number of replicates cannot be directly quantified without experimental data. The minimum number of replicates required depends on a number of experimental factors, including absolute signal intensity, signal to noise, and microarray to microarray variability. Experiments should be set up to provide a p-value of <0.01 for statistically significant data.
Why is it important to use the log transform of my raw ratio values before computing most statistics?
The Speed Group Microarray Page* at the University of California, Berkeley provides a good explanation for the importance of using the log transform of raw ratio values before computing most of your statistics.
*Note: Some of the links on this page connect to information sources outside of NIAID and are provided as a convenience for World Wide Web users. Please see the NIAID disclaimer.
back to top