Hi,
I have done a microarray and I have my data in a txt file (probes and intensity) This file contains data from around 50K probes and two conditions I have already analysed this data and no significant results have been found.
I’m interested in study genes related with cell differentiation between my two conditions. Is it correct to only take probes related with cell differentiated genes from my txt file and re-analyse my data? Since I will have fewer probes perhaps some genes will pass FDR correction.
If this way is not correct, with method do you recommend to study cell differentiation genes between my conditions?
Thank you
1 answer
Is it correct to only take probes related with cell differentiated genes from my txt file and re-analyse my data? Since I will have fewer probes perhaps some genes will pass FDR correction.
Assuming you did your first analysis correctly, then no, it is not correct to select a subset and try again, even if it makes sense with the biology of your question - this is a form of P-hacking.
What you could try is checking for concerted gene changes with GSEA or some similar method, which might be picked up where individual genes analysis was not too sensitive.
Log in to answer this question.
-
Proteomics normalization: equal protein loading but unequal cell counts in clinical samples
written by Kittinun •
-
ScRNAseq-How to correctly choose cell type marker genes
written by Francesco •
-
Is bisulfite-seq data from different studies comparable with each other?
written by CTLong •
-
How to investigate correlation between transcript and protein abundance?
written by Wang Cong •
-
Differential Gene Expression after batch correction with DESC?
written by skjw1029 •
-
genes not mapping to ENTREZID
written by adR
-
Interpreting a list of Differentially Expessed Genes
written by ajp104 •
-
How to analyse TXT format microarray data
written by afli
-
Script for analyzing idat (illumina) microarray files with limma, any suggestion?
written by Pol
-
Doesn'T It Make Sense To Reduce Multiple Testing By Being More Selective With Genes Investigated?
written by jobinv
If you have done the analysis with the data and if you have not found any differentially expressed genes, there ain't much to do. You can plot expression of the set of genes (barplot, heatmap) that you are interested in but they wont show significant change as expected. I don't think you can perform the normalisation with a small set of genes. That will produce erroneous results.
Thank you for your answer.
Take a look at these previous threads: Doesn'T It Make Sense To Reduce Multiple Testing By Being More Selective With Genes Investigated? and Is it OK to select a small subset of expression data for statistical testing?
Thank you for your answer.