calculating logfold change from variance stabilizing transformation
1
0
Entering edit mode
@lirongrossmann-13938
Last seen 12 months ago

Hi All,

I am interested in calculating the logfold change from the transformed expression file obtained by using variance stabilizing transformation function. 

I used the transformed data to normalize my count data and compare two groups within the data using a different statistical test than the one in Deseq, thus I cannot use the logfold change given by the differential analysis of Deseq. 

Is there a straightforward conversion between the logfold change and the values obtained by variance stabilizing transformation?  (I tried taking the exp of the difference between the averages of the two groups, but it did not work).

thanks 

deseq2 logfoldchange variancestabilizingtransformation • 611 views
ADD COMMENT
1
Entering edit mode
@mikelove
Last seen 13 hours ago
United States

Transformed data is on the log2 scale for large counts. 

Note that we recommend use of DESeq, results, and lfcShrink for the optimal estimation of log fold changes.

ADD COMMENT
0
Entering edit mode

My groups are small (9 vs 9 samples) and for some reason when I tried a different statistical test to compare them (Wilcoxon rank sum test, to be specific), I got a more uniform expression within each group for the genes that were significant. When I used Deseq to perform the statistical test, the genes that were found to be differentially expression had a large variance within each groups. 

I would really like to use Deseq for my statistical analysis as well, is there a way to overcome this problem?

ADD REPLY
0
Entering edit mode

I’d recommend using a package specifically designed for the data, as it will have known operating characteristics, whereas custom solutions are just less known quantities. So if you find that you prefer working with ranks for this dataset, I’d recommend using SAMseq.

ADD REPLY
0
Entering edit mode

Thank you so much! 

I see that SamSeq requires normalized data to run. From your experience, is it “safe” to give SamSeq the output of variance stabilizing transformation from Deseq2 as the input for the analysis.

ADD REPLY
1
Entering edit mode

You should provide SAMseq with the raw counts as 'x':

https://www.rdocumentation.org/packages/samr/versions/2.0/topics/SAMseq

It uses its own resampling strategy in lieu of parametric normalization. It works well.

ADD REPLY

Login before adding your answer.

Traffic: 170 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