choose normalized counts or log2FC when do correlation in RNAseq

Question

[dew]

Dear guys,

Which one is better/prefers to do correlation analysis for the RNAseq samples? Use the normalized counts or log2FC(comparison groups)? If use the log2FC to do the correlation of different comparisons, how to deal with the pvalue/padj of each log2FC?

Thank you very much!

Answer 1

[bioinfo_ga]

Normalized counts are always better to identify the samples correlating and the outliers. log2FC is already a single value for two or more samples and will not give an idea of individual samples. pvalue/padj should only be considered if you have replicates in your study, preferably 3 or more .