Question: What is the significance threshold for differential gene expression (10folds, 50folds, 100folds) between tumor and normal sample?
0
gravatar for bioinforesearchquestions
2.9 years ago by
United States
bioinforesearchquestions250 wrote:

Dear All,

I am working on the RNAseq data. I would like to know what is the significance threshold for differential gene expression (10folds, 50folds, 100folds) between tumor and normal sample?

rna-seq • 1.5k views
ADD COMMENTlink modified 2.9 years ago by WouterDeCoster40k • written 2.9 years ago by bioinforesearchquestions250
1
gravatar for WouterDeCoster
2.9 years ago by
Belgium
WouterDeCoster40k wrote:

The significance is the (adjusted) p-value, not the fold change. Although it's common to set a cut-off on the (log) fold change there is not really a need to do so. An argument in favor of filtering on fold change could be that very small effects (fold changes) are unlikely to produce large biological effects, but I would argue against that.

ADD COMMENTlink written 2.9 years ago by WouterDeCoster40k

This is precisely the point, one should not figuratively filter on log2(FC) values first. First value of filtering is your FDR corrected pvalues and this conserves the fact os 1% or 5% error rate. After that if you still see too many genes then you can filter out low expressed genes that have very might fold change unless they are not having miRNA as sometimes small changes might also induce some phenotypic changes or induce some pathways that might infer the phenotype. Usually in cancers the cut off is first at adj.pvalue of either 0.01 or 0.05 depending upon the number of DEGs and then still if you have a lot of differentially expressed genes then filter out genes with low fold changes as they might not give large biological effects which can range from 1-1.5 log2(FC) value for cleansing post adj.pvalue filtration to give much more robust gene sets.

ADD REPLYlink written 2.9 years ago by ivivek_ngs4.8k

Thanks, WouterDecoster and vchris_ngs.

I am pretty much new to this RNAseq. I have the following cuffdiff output files, - gene_exp.diff - genes.fpkm_tracking - genes.count_tracking - genes.read_group_tracking

similarly for isoforms, cds and tss group.

Can you please shed some light on carrying out these steps? Do you have any research articles discussing it in detail?

ADD REPLYlink modified 2.9 years ago • written 2.9 years ago by bioinforesearchquestions250
1

Perhaps it's worth trying this very recently published workflow: http://www.ncbi.nlm.nih.gov/pubmed/27560171

ADD REPLYlink written 2.9 years ago by WouterDeCoster40k

Thanks, I will go through it.

ADD REPLYlink written 2.9 years ago by bioinforesearchquestions250
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1325 users visited in the last hour