Question: Is fold change of value 1.5 (log2FC = 0.58) a significant value ?
gravatar for anandprem1792
21 months ago by
anandprem179220 wrote:

In Biostars forum for differential gene expression analysis by LIMMA I found that most people suggest to set a threshold value of Log2FC > or = 2 to filter DEGs. When I analyzed the GEO dataset "GSE90594" - Agilent-039494 SurePrint G3 Human GE v2 8x60K Microarray 039381 platform for DEGs by LIMMA package, I found that very few genes only have a log 2FC value above 1.

But in the original paper "Study of gene expression alteration in male androgenetic alopecia: evidence of predominant molecular signalling pathways" (, the authors reported that they obtained 325 UP and 390 DOWN regulated DEGs. They have followed a different approach for obtaining DEGs and reported the DEGs with Fold change (FC) values. Most of the UP regulated genes are reported with the FC value 1.5 to 2 range (log2FC value 0.58 to 1.3 approx). Similarly for down regulated genes FC values of 0.5 to 0.8 (log2Fc value -1 to -0.3) are considered.

Is this right ? can we consider log2FC value of above 0.6 for UP and below -0.6 for DOWN DEGs. Please someone clarify this irrespective of p and FDR values. I consider a q value of 0.05 for DEG selection.

ADD COMMENTlink modified 21 months ago by WouterDeCoster45k • written 21 months ago by anandprem179220
gravatar for Devon Ryan
21 months ago by
Devon Ryan98k
Freiburg, Germany
Devon Ryan98k wrote:

Filter first by adjusted p-value and only then by a fold-change that makes sense for your experiment. This might be a fold-change of 1 or 0.5 or 2 or something else, there's no one-size-fits-all solution to that. In some cases you may not filter by fold-change at all.

ADD COMMENTlink written 21 months ago by Devon Ryan98k

Thank you for the suggestion. As you said I will filter by "adjusted p-value" first.

ADD REPLYlink written 21 months ago by anandprem179220
gravatar for WouterDeCoster
21 months ago by
WouterDeCoster45k wrote:

The fold change will tell you something of the size of the effect of differential expression, so it's more about biology than about statistics. If you are looking for genes with big changes, you'll pick a higher cut-off. But if a gene is significantly differentially expressed then it's already worth looking into it: subtle differences in gene expression can already have a substantial impact. It probably makes a big difference in which gene is affected, too. Some genes are more dosage-sensitive than others.

ADD COMMENTlink written 21 months ago by WouterDeCoster45k

Thank you for the help.

ADD REPLYlink written 21 months ago by anandprem179220
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 2104 users visited in the last hour