normalize on chip-seq data

Question

[catherine]

I have two replicates of chip-seq data. And I run bowtie and call peaks using MACS separately.

MACS2 callpeak -t R1.bam -c Input1.bam -f BAM -g dm -n macs2_R1 -s 50 --call-summits --bw 151
MACS2 callpeak -t R2.bam -c Input2.bam -f BAM -g dm -n macs2_R2 -s 50 --call-summits --bw 149

Now I want to combine them together for further analysis.

Do I need to normalize them before combine together? I put input when running MACS, I thought MACS has done normalization when calling peaks.

If I should do normalization firstly, and people always say "Normalization by sequencing depth (i.e. total read count)" or "divided by total reads". How to do this? I don't have "total reads" after running thru MACS.

How can I do to combine these two sets of peaks?

Thank you!!

[Manvendra Singh]

May be you are looking for algorithms like MAnorm

http://genomebiology.com/2012/13/3/r16

[catherine]

I tried to use MAnorm, but I couldn't install the pre-requesite packages for MAnorm : R.basic and MASS. I used:

source("http://www.braju.com/R/hbLite.R")                                                                                                                                                    
hbLite(c("R.basic","MASS"))

[Qi Zhao]

MAnorm is a good choice.

catherine12243,

R.basic package has already been deprecated. From http://www.braju.com/R/

Many of these functions have now been moved to R.utils and aroma.light.

You can install R.utils and aroma.light instead by following command:

source("http://bioconductor.org/biocLite.R")
biocLite("aroma.light")
install.packages(c("R.oo","R.utils","MASS"))

Answer 1

[Ming Tommy Tang]

1. You can call peaks separately and then use the common peaks as final peaks.
2. You can also just merge the bam files by samtools merge (See this post) and then used the merged bam file and the input bam files for calling peaks. MACS should do the depth normalization by itself.
3. You have replicates, try pepr and multiGPS

[catherine]

If I just use the common peaks (your method1), is the summit height reliable?

Answer 2

[Qi Zhao]

MAnorm is a good choice.

catherine,

R.basic package has already been deprecated; "Many of these functions have now been moved to R.utils and aroma.light." posted from http://www.braju.com/R/

You can install R.utils and aroma.light instead by following command:

source("http://bioconductor.org/biocLite.R")
biocLite("aroma.light")

install.packages(c("R.oo","R.utils","MASS"))