Question: Microarray analysis - hill shaped p value histogram
gravatar for ulapei
15 months ago by
ulapei20 wrote:


I am trying to analyze Affy microarray data. For this, I used limma package and I noticed that I get unexpected results aka none of the genes come out as significant. After checking the p value histograms, I noticed an unusual trend - histograms appeared to be hill shaped: I have done PCA and this has indicated two possible outliers but even if I remove them I still obtain the plot that is the same shape.

I assume this indicates that the wrong test was used. Does anyone know how I could possibly make the p plots appear to have a uniform distribution? Is there a different test that is used in this case?

Thank you

ADD COMMENTlink modified 15 months ago • written 15 months ago by ulapei20

that distribution is as uniform as I've seen in a microarray analysis. You don't typically _want_ a uniform distribution - you'd expect a uniform distribution if there was no differences between your groups; if there is a detectable difference, you normally see a peak of features with very low p-value.

Is there any evidence in the literature that specific genes are differentially expressed under the conditions studied? If so, I'd have a look at those genes within your dataset.

Are there confounding, or batch/inter-individual effects that you haven't accounted for? If so control for these factors and reanalyse your data

Is the biological question really subtle? If so, even if differentially expressed genes existed you might not have enough power to detect changes in those genes.

ADD REPLYlink written 15 months ago by russhh4.8k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1460 users visited in the last hour