Normalizing small RNA Seq Reads

Question

[empyrean999]

Hello All,

This could be a easy solution but just wanted to get your opinion on this. We have 10 different small RNA samples sequenced for arabidopsis. They are varied in read numbers from 10 million to 30 million . We wanted to look at reads mapping to 5' end of a gene. I used that gene as reference and mapped all the reads and got read counts mapping to it.

So to get relative expression, can i use simple normalization using total reads as my reference is only one gene or should i use proper techniques such as TPM or TMM ?

[Martombo]

what is exactly your goal? assessing differences in the 5'UTR coverage of that gene?

[empyrean999]

The goal is to understand difference in num of reads mapping to that 5' region of gene approx (150bp). Reads only map to that part and no reads seen in rest of the gene. I have read counts only for that 150bp location . I wanted to see across different samples, what are the normalized read numbers?

For example

Sample name   Total No of reads   Reads mapped   Normalized num
1             100                 5              5/100 = 0.05
2             200                 20             0.1
3             1000                50             0.05

Would the above normalization be enough in my case or is there anything I am missing?

[Martombo]

you said you have small RNA samples right? if so, you would actually de-rich mRNA sequences in the process (even with size selection only) and for this reason, looking at a small window of a gene would not give you a reliable indication of a biological process. or are you actually making the hypothesis that the reads you see on the gene would come from a small RNA that would target the gene?