Z-Score On Raw Counts

Question

[camelbbs]

Hi All,

I have a question on normalization of rnaseq data. After I get raw reads counts from htseq-count, I want to use zscore to normalize it. Is that reasonable?

And how I can do that. If there have some guides, that will be very helpful.

Thanks a lot!

che

[Steve Lianoglou]

Can you specify why you want to z-transform the data? What are you planning to do with it? My first guess is that this is not a reasonable thing to do, but please provide more information all the same.

[camelbbs]

I want to make heatmap. And Zscore can make the values centralized by 0, right?

[Damian Kao]

You need to normalize your raw count data first. Z-score will standardize your data, but it won't normalize by library size. Another option is to use variant stabilization method provided by DESeq.

[Sean Davis]

What Damian suggests is what you want....

[camelbbs]

Hi Damian, Thanks very much. DO you mean Zscore is not a normalization method? IS there any common method for normalizing the raw count, except for DESeq... Can I calculate RPKM first and then do zscore for rpkm...

[camelbbs]

I also want to ask how to use zscore to standardize my data, is there any package in R?

[Sean Davis]

Standardization can be done using the sweep() function in R.

[camelbbs]

Thanks a lot Sean.