Question: why expression value must be log2 transformed before further analysis, such differential analysis?
gravatar for jlshi.nudt
6.4 years ago by
jlshi.nudt210 wrote:

Hi, just as the title.

When using microarray, the intensity should be log transformed, that is partly because the intensity values are relative numbers. But in RNA-Seq, why must RPKM be log transformed?


I also notice that the differential analysis results are largely different with/without log transform. why? anyone tell me? thank you.

expression log • 23k views
ADD COMMENTlink modified 6.4 years ago by mikhail.shugay3.4k • written 6.4 years ago by jlshi.nudt210

Who told you this? Log transformed expression are easier to detect DEG with low expression level. However, I never heard it's a must to log transformed them.

ADD REPLYlink modified 9 months ago by RamRS30k • written 6.4 years ago by Xingyu Yang260

I just saw this transformation on many papers resolved with RPKM and microarray expression data. It seems to be a usual convention.

Btw, I guess that low expression levels are not believable, so I always to filter with some threshold value.

ADD REPLYlink written 6.4 years ago by jlshi.nudt210
gravatar for Devon Ryan
6.4 years ago by
Devon Ryan97k
Freiburg, Germany
Devon Ryan97k wrote:

The general reason to log-transform data (log2 or otherwise) is to make variation similar across orders of magnitude. This isn't really a must, but usually makes things more convenient. Having said that, tools like limma are expecting log2 values, so if you're going to plug your RPKMs into it then it'd be a very good idea to log2 them first.

ADD COMMENTlink written 6.4 years ago by Devon Ryan97k
gravatar for mikhail.shugay
6.4 years ago by
Czech Republic, Brno, CEITEC
mikhail.shugay3.4k wrote:

Indeed microarray values and RPKM/FPKM values are better correlated when log-transformed. The reason for it is that the distribution of RPKM/FPKM values is skewed, and by log-transforming it we could bring it closer to normal distribution. It is needless to say that many statistical tests require normally-distributed data..

ADD COMMENTlink written 6.4 years ago by mikhail.shugay3.4k
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: 2202 users visited in the last hour