How to properly use and understand chip-seq analysis with DiffBind ?
0
0
Entering edit mode
@pierremarin-17659
Last seen 8 months ago

Hi everybody !

I'm a PhD student trying to analyse Chip-seq data generated in my project but I'm loosing it to use properly DiffBind package even if I read several times the manual.

To illustrate my work :

Chip-seq on flies:

• 3 Lineages (A, B and C) x 3 treatments control included (CT, PQ, CO) x 2 histones mark K4 and K9 x 2 replicas per conditions (ov1 and ov2)

= 36 samples analyzed, 18 for K4 and 18 for K9.

My questions are :

1. Effect of treatments per lineage ?:

In lineage A i want to contrast control (CT) against CO or PQ and the same thing for other lineages

1. Effect of lineages per treatment? :

In treatment CO, difference between A and B , B and C...

If I understand the Diffbind package usage :

1. I imported data in a csv format with .bam file for sequence informations, and peak caller data obtained from Peak ranger software
 ID TISSUE FACTOR CONDITION REPLICATE INTERVALS A_CT_k4_ov1 A K4 CT 1 A_CT_k4_ov2 A K4 CT 2 A_CO_k4_ov1 A K4 CT 1 A_CO_k4_ov2 A K4 CT 2

I used to import :

samples_k4 <- read.table(data-sheet)
chip_k4 <- dba(sampleSheet =samples_k4)


Then I used the count option which count peaks in the peak caller datasif I understand but I didn't understand if it use information of input data given

chip_k4_count <- dba.count(chip_k4, summits = 250, bParallel = F)


After counting I used the option to generate contrast

chip_k4_contrast <- dba.contrast(chip_k4_count,  categories = c(DBA_CONDITION, DBA_TISSUE)


To show contrast possibility I used

dba.show(chip_k4_contrast, bcontrast = T)


And now problems...

For example the contrast between lineages, group1 = A vs group2 = B take all sample of the lineage A and B so control and other treatment together...

But I want group1 = A:CT vs group2= B:CT How to use a fixed factor as the treatment, I tried with block option without success.

How work Diffbind with replicas ? Because I want to made Ven Graph after and when i tried, I obtained graph with specific replicas and no a global effect.

Thanks for every effort to help me !

diffbind chip-seq R • 390 views