log2 fold change in result of DEseq2
2
0
Entering edit mode
@elhamdallalbashi-11418
Last seen 4.9 years ago

Hello,

I have a question about influence of log2 fold change in determining best genes in differential expression of DEseq2,as an example with adjusted p-value < 0.1,I have 23 genes,

out of 26998 with nonzero total read count

adjusted p-value < 0.1

LFC > 0 (up) : 10, 0.037%

LFC < 0 (down) : 13, 0.048%

outliers [1] : 241, 0.89%

low counts [2] : 11375, 42%

(mean count < 1)

what is the best range of log2 fold change for selecting genes?

deseq2 • 1.7k views
ADD COMMENT
0
Entering edit mode
theobroma22 ▴ 10
@theobroma22-11920
Last seen 4.9 years ago

Question: did you have technical replicates?? Those stats are quite odd. 

ADD COMMENT
0
Entering edit mode

The numbers in themselves don't necessarily imply too few biological replicates. The numbers you detect depend on numerous properties of the experiment or study, and the conditions or treatments being compared (or more complex relationship for more complex experimental designs): as you mention the sample size (number of biological replicates), but also the true effect sizes or interaction sizes for each gene, the variability among biological replicates, the sequencing depth, just to mention a few.

It could be an under-powered experiment, or an under-sequenced experiment, or a well-powered experiment with sufficient sequencing depth, low biological variability and the "treatment" only induces a few changes in expression. 

ADD REPLY
0
Entering edit mode
@mikelove
Last seen 3 hours ago
United States

Can you be more specific about exactly what you are interested in? You could choose the genes with lowest adjusted p-value, or you could rank the genes among the significant set by the absolute value of the log2 fold change.

ADD COMMENT
0
Entering edit mode

Hello Michael,

if you remember, I wanted to do DE for all treated vs all controls samples.and you suggested me a design of ~experiment + condition, I need DE for co-expression network,so for coding and non coding I did that,but I see that there are a variety value in log2 fold change.I can not understand which has important role in select genes,fold change or adjust p-value?

this the result for coding;

out of 19933 with nonzero total read count

adjusted p-value < 0.1

LFC > 0 (up) : 335, 1.7%

LFC < 0 (down) : 300, 1.5%

outliers [1] : 289, 1.4%

low counts [2] : 6012, 30%

(mean count < 61)

I should say that samples are similar for coding and non coding but the result are very different!

ADD REPLY
0
Entering edit mode

"I can not understand which has important role in select genes"

I can't help you with this one, this is up to you as the analyst. You might discuss with a local statistician the difference between effect size (log fold change) and statistical significance (p-values, FDR sets), and which is more appropriate for your analysis.

 

ADD REPLY

Login before adding your answer.

Traffic: 285 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6