Question about normalization and differential expression analysis

Question

[kamel]

Hello,

I study the diferential expression between two samples of two different conditions, I made the mapping of my fatsq (single-end) files on the reference genome and then quantified by Feature-counts. I would like to know what is the good method of the normalization method (RPKM, TPM ...) and by which package do you advise me to use for differential expression analysis ??

Thank you.

Answer 1

[grant.hovhannisyan]

How many samples per condition do you have? If at least two (though in general minimum 3 is recommended) then you can use bioconductor package DESeq2, it is one of standards in DE analysis, which performs its own normalization procedures (you have to feed it with raw read counts which you got from featurecounts).

[kamel]

I have only one sample mRNA per condition, so I have only two samples, you find below a preview of my counting file better that I use deseq2 with that propore method of normalization ???

Geneid  Chr Start   End Strand  Length  sample1.bam sample2.bam
NMPOQ_01G00001-E1   1   10466   11000   -   535  1  2
NMPOQ_01G00001-E2   1   8658    10408   -   175  1  7   
NMPOQ_01G00001-E3   1   8595    8609    -   15   0  0
NMPOQ_01G00001-E4   1   8116    8523    -   408  0  0
NMPOQ_01G00001-E5   1   7727    8063    -   337  0  5
NMPOQ_01G00001-E6   1   7417    7676    -   260  0  1
NMPOQ_01G00001-E7   1   7041    7358    -   318  0  9
NMPOQ_01G00001-E8   1   6865    6972    -   108  0  2
NMPOQ_01G00001-E9   1   6728    6812    -   85   0  2
NMPOQ_01G00001-E10  1   6365    6676    -   312  1  4
NMPOQ_01G00001-E11  1   6089    6096    -   8    0  0
NMPOQ_01G00003-E1   1   12130   12168   -   39   0  0
NMPOQ_01G00003-E2   1   11793   12083   -   291  0  0

NMPOQ_01G00003-E3 1 11702 11736 - 35 0 0 NMPOQ_01G00003-E4 1 11394 11649 - 256 0 0## Heading ##

[grant.hovhannisyan]

Unfortunately with this experimental set-up (one sample per condition) you can not meaningfully answer your biological question (differential expression between these two samples). What you can do is some exploratory analysis with your data. Here are some threads where you can find more info on this topic

Advice on RNASeq analysis without biological replicates for differential gene expression.

http://seqanswers.com/forums/showthread.php?t=31036

https://support.bioconductor.org/p/61531/

Answer 2

[lakhujanivijay]

I would like to know what is the good method of the normalization method (RPKM, TPM ...)

That depends entirely on your needs. A quick reference to understand the differences could be found here and here

and by which package do you advise me to use for differential expression analysis ??

I will suggest going with the new tuxedo protocol. A quick guide is here

[kamel]

I already saw these links, thank you in any case. i want a good method in my case and as i said i study the differential expression between two sample mRNA (single end)