normalization of pair-end RNA-seq data with different sequence length

[yao_licr]

Hi all,
I have downloaded 150 pair-end RNA-seq data (patient samples). There are several different sequence lengths of these RNA-seq data, from 50 bp, 75 bp, 100 bp, 101 bp to 103 bp. Moreover, the library sizes also differ quite a lot (from around 20 million to 110 million reads). I would like to divide these 150 samples into two groups, 1p36 positive or negative, based on the expression level of genes located at the chromosome region 1p36. So the problem is how to normalize this RNA-seq dataset. Any suggestions?

Thanks a lot!

Yao

[bastianwur]

I don't think the differences in read length should matter too much (practical example: People do normalization with trimmed reads, they all have different lengths).
Because what you want is the frequency, in which a mRNA appears in a cell. How long the mRNA has been sequenced is not relevant for this.
(AFAIK)