Data sets suitable for analysing gene expression

Question

[fawazfebin]

Hi

I was analysing some drug resistant cancer data sets ( RNA - Seq) for determining major genes which are differentially expressed. I already analysed an experimental data set from GEO . Inorder to arrive at reliable results to get published how many samples should I select for the analysis? Should the samples be from different patients or from cell lines? And from where I could get suitable data sets? I would also like to perform a multi omic analysis with GSOA. And would like to know a preferable source for the datasets to perform GSOA. Great thanks in advance.

[russhh]

use all the samples that are available.

Answer 1

[Kevin Blighe]

Inorder to arrive at reliable results to get published how many samples should I select for the analysis?

You could perform some power analysis. However, for RNA-seq, due to the fact that we are looking at transcribed mRNA and not just variants (which may not even alter mRNA transcription), you can typically get away with low sample numbers. This will, however, greatly affect where you can publish the work.

Should the samples be from different patients or from cell lines?

This is a decision for a biologist, not a bioinformatician. If you want patient data, then they should be matched on various clinical features and not just a 'random' bunch of people. Match on as many clinical and experimental factors as possible. Same for cell lines.

And from where I could get suitable data sets? I would also like to perform a multi omic analysis with GSOA. And would like to know a preferable source for the datasets to perform GSOA. Great thanks in advance.

For ideal repositories, take a look at Recommended Data Repositories