--Hi all,

I am struggling with a statistical question related to RNA seq data. I have collected cells sets and then extract RNA data from 6 different persons. I have 2 conditions: treated and untreated. At the end i have 12 libraries: 6 treated and 6 untreated. My goal is to find genes differentially expressed between the conditions. After recieving back the sequencing data (2x50bp illumina paired-end sequencing) I noticed a clear batch/individual effect between the 6 sets of samples because genes are higher (additional variation in the counts) in some persons as compared to the other.

I have used DEseq2 package: first without using batch factor and then with batch factor, I have more differentially expressed genes including batch factor.

So, do you think I should build my model including the batch factor (different donors) to the design to increase the sensitivity for finding differences due to condition or not ?

Any feed back on possible alternatives methods I can use are very much appreciated!

thank you so much.

Regards, Laurent --

Here is my sampleTable:

  sampleName       fileName condition    batch
1    treated1.dat   treated1.dat   treated    D1
2    treated2.dat   treated2.dat   treated    D2
3    treated3.dat   treated3.dat   treated    D3
4    treated4.dat   treated4.dat   treated    D4
5    treated5.dat   treated5.dat   treated    D5
6    treated6.dat   treated6.dat   treated    D6
7  untreated1.dat untreated1.dat untreated    D1
8  untreated2.dat untreated2.dat untreated    D2
9  untreated3.dat untreated3.dat untreated    D3
10 untreated4.dat untreated4.dat untreated    D4
11 untreated5.dat untreated5.dat untreated    D5
12 untreated6.dat untreated6.dat untreated    D6

See the workflow:

https://bioconductor.org/packages/release/workflows/html/rnaseqGene.html

We use an example with four individuals and untreated and treated samples for each individual.