With Luminex multiplex data, there can be a large number of analytes assessed in a single assay or across multiple assays. In regards to the statistical analysis of the data. I have seen authors publish their analysis as essentially treating each assay individually, using a t-test or ANOVA to analyze their data. My question is, do you need to statistically correct for the number of assays you're assessing or do you not have to correct?
Let's say, for example, I run 30 analytes and I find 2 analytes to be particularly interesting that have obtained significance by treating each assay as it's own event without correcting for the 30 assays examined. Is it okay to report that data as significant? Or do I need to correct for the 30 total assays that were run? As a follow-up question, if I re-run the interesting analytes that obtained significance on a totally separate independent assay and obtain significance there, would that be acceptable to report?
I'm wondering because if you look at a number of assays and you need to statistically correct for the quantity you're looking at, you essentially will lose all significance because the p-value for obtaining significance becomes absurdly small for what is essentially ELISA data.
Thanks for your help!
Hi Edwards
I could not follow your statement that ' you essentially will lose all significance because the p-value for obtaining significance becomes absurdly small for what is essentially ELISA data'.
Will u pl.check your statement?
I do agree with the answer given by Ajit Kumar Roy.
I see the statement to be confusing.
If you analyze each assay as it's own assay, the p-value is 0.05. If you have 30 assays, you would divide the p-value by the number assays, which is 0.05/30 = 0.00167. So if you have an analyte that when analyzed individually had a p-value = 0.01, that would then not be considered significant if you correct for the number of analytes examined.
I'm not sure given the nature of the technique if this is necessary to correct for the number of analytes, or if it is acceptable to consider them individual assays. Because if you run 30 analytes you would expect at least 1 to be significant by chance.
- Badges
- Science topic
- Similar topics
- Microbiology
- Assays