bamCoverage normalization for Chip-seq

[garen...@gmail.com]

Hi,

Recently, I tried to compare several Chip-seq data from different platform, and I thought deeptools maybe helpful for me.

According to published paper: "For each chromatin mark and the input control, we divided the genome into 100 bp bins, counted the number of reads falling into each bin i, and computed the number of reads per kilobase of bin per million reads sequenced (denoted RPKMmark,i), with the exception of the input where RPKM is computed over the five consecutive bins centered at i. Because input represents the entirety of the genome and is not sequenced to saturation like ChIP data, a 500 bp bin is used to reduce noise that occurs in a 100 bp window. Finally, normalized ChIP enrichment is computed as DRPKM = RPKMmark,i – RPKMinput,i." I am trying to use bamCoverage to do this normalization, however, I am confused these parameters. I tried followed command:

bamCoverage --binsize 100 --bam chip_sum_sorted.bam --normalizeUsing RPKM -o chip_sum.bw

I am not sure whether this command worked or not? Could you give me some idea?

Thank you so much!
Best,
Garen

[Devon Ryan]

That's fine for the ChIP signal, for the input signal use
`--smoothLength 2500 --binsize 500` to match that paper. I have to
say, however, that I don't like what that paper is doing (not least of
which because comparing RPKM scores where different bin sizes were
used can lead to odd results). The better route is to take the BAM
files is run `bamCompare` on them (usually with the default log2 ratio
as the output, though you can subtract the signals instead if you
prefer.

Devon
--
Devon Ryan, Ph.D.
Email: dpr...@dpryan.com
Data Manager/Bioinformatician
Max Planck Institute of Immunobiology and Epigenetics
Stübeweg 51
79108 Freiburg
Germany

> --
> You received this message because you are subscribed to the Google Groups "deepTools" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to deeptools+...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[garen...@gmail.com]

在 2018年7月6日星期五 UTC+8下午7:20:08，Devon Ryan写道：

Dear Devon,

Thank you for your prompt reply. I will try the "bamCompare", because I used different depth of download data, so which normalization method is better? (SES or RPKM?)

Thank you!
Best,
Garen

[Devon Ryan]

SES for signals that aren't broad, otherwise I prefer RPGC (aka, 1X)
normalization.

Devon
--
Devon Ryan, Ph.D.
Email: dpr...@dpryan.com
Data Manager/Bioinformatician
Max Planck Institute of Immunobiology and Epigenetics
Stübeweg 51
79108 Freiburg
Germany