ChIPseq: clonal reads or peaks?

[Ohad]

Can anyone offer an idea ?

[Ohad]

I`m working too these days on ChIP data using HOMER and I encountered a related problem.

HOMER preform Basic Quality Analysis (attached) when executing makeTagDirectory which produces tagCountDistribution.txt file.

You should view this file on EXCEL. As far as I read HOMER does not filter out clonal reads but simply inform you they exists on your data. If your library is "too clonal" you can try to add the -tbp <#> option to makeTagDirectory command and limited the # of reads allowed per position.

Regarding tagCountDistribution.txt my understanding says that a line which looks like:

2 (tab) 0.11

represents that 11% of the reads are the same. Meaning , according to homer, that the read's first base has the same annotation. (I personaly removed all non-unique reads with samtools view -q 30)

I attached my tagCountDistribution.txt and I see on the first line:
0 (tab) 0.716642
I cannot understand what this line represents. How can a mapped read fall on "0" position or "no-position" ?

Does anyone understand their documentation better ?

*You can find more info on:
http://homer.salk.edu/homer/chipseq/qc.html

[crocky]

Hi guys,
I have tried to analyze my ChIPseq data using Homer. No when I look at some called peaks via IGV, it seems that some peaks consist of clonal reads, though Homer should filter out these peaks, shouldn't it? I have attached a snapshot, so that one can see, what I mean (black is input, yellow is IP1 replicate 1, green is IP1 replicate2, red is IP2).
For mapping I have used bowtie2 with --local. Could it therefore be, that these reads were trimmed resulting in such "sharp" peaks? Furthermore I have used dm3 as reference genome, since it was pre installed in Homer, and is also the one used in UCSC. Could this older genome version affect the mapping so badly? Does anyone know whether these are true peaks , or rather artifacts? Interestingly these peaks do not come up in input control, but only in IP against my transcription factor of interest.

Thanks for your help,
Claudia