I have got a question about data. I obtained Differential Expression Log2 Ratios of genes with Differential Expression Adjusted p-values. I removed those whose p-value was higher or equal to 0,05. I want to focus on some genes for 4 treatments. Unfortunately, in one of four treatments in one gene, the p-value is higher than 0,05. So my question is if I can I use it for heatmaps creation or not?
Thank you for your answers.
P-values selection doesn't usually work for much of anything . See the case in regression in the attachment. Best wishes David Booth
Yes, you can certainly use that to make a heatmap.
I would suggest that you select the genes to be shown on the heatmap by the ANOVA. Decide on how many genes you like to show (="n") and use the "top-n" genes according to the ANOVA F- or p-value.
I love seeing how simple words as "bad" and "unfortunately' attached to pvalues can still fuel a century old bias, like pvalues carried any responsibility and researcher had to always show small pvalues 🤷♂️
Anyways, I would probably add one thing to what jochen said:
I would not test "zero" but a more biological meaningful value as it is usually done in practice and use the original pvalues ( with null hypothesis being the absolute logFC larger than 1 or 2 or whatever you deem reasonable ) and rank the genes in order for you to focus on a handful of genes
Dear Stefano Nembrini , I suggested to use the ANOVA statistics (F-values or, if the effective sample sizes between genes are different, p-values), not p-values from test of individual logFCs as the statistik to rank order the genes. Your "addition" to my suggestion is referring to individual logFCs, what is incompatible with my suggestion.
But if one is using tests on individual logFCs, then your hint that a more biologically meaningful null hypothesis should be tested is very sensible. I certainly agree with that.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA