[Bioc-sig-seq] ChIP-seq analysis in normalization/peak calling between sample and control

[Muino, Jose]

Hi Charlie,

"Nhits" indicates the number of locations where the read can be aligned. This value is used by mappedReads2Nhits function when parameter uniquelyMapped=TRUE to eliminate these reads with a Nhits value different of 1.


-----Mensaje original-----
De: bioc-sig-sequencing-bounces at r-project.org en nombre de Chen-Yi Chen
Enviado el: lun 29/03/2010 23:52
Para: Muino, Jose
CC: bioc-sig-sequencing at r-project.org
Asunto: Re: [Bioc-sig-seq] ChIP-seq analysis in normalization/peak calling between sample and control
 
Hi Jose,
I tried to use our AlignedRead object feeds into mappedReads2Nhits function from CSAR library, but it returned with an error:
> ChIPnhits = mappedReads2Nhits(ChIPRead, "ChIPnhits_test", chr, chrL)
Error in input$Nhits : $ operator not defined for this S4 class

I believe we don't have "Nhits" column when we read the aligned output from ShortRead, since we are using Bowtie alignment. So here's my question, what does "Nhits" in here means? I believed I need to know what this mean in order to do a custom format read-in for "loadMappedReads."

Thanks!

-Charlie-


----- Original Message -----
From: "Muino, Jose" <jose.muino at wur.nl>
Date: Saturday, March 27, 2010 2:43 am
Subject: RE: [Bioc-sig-seq] ChIP-seq analysis in normalization/peak calling between sample and control
To: Chen-Yi Chen <ChenYiChen at lbl.gov>, bioc-sig-sequencing at r-project.org

> Hi Charlie,
> CSAR package has their own function to load the mapped read files  
> (loadMappedReads).  Although, it is also compatible with the class 
> AlignedRead from  ShortRead package;  you can directly use an 
> AlignedRead object as input on the mappedReads2Nhits function. Let 
> me know if it doesn´t work for you
> 
> Most likely the problem with the output wig file is that the 
> chromosome names used by the genome browser (eg: chr1,chr2.) is 
> different to the  chromosome IDs of the fasta file that you were 
> using to map the reads. 
> If you are using the last version (0.99.4), the easy way to change 
> chromosome names is in the output of ChIPseqScore. Eg:
> R> test <- ChIPseqScore(control = nhitsC, sample = nhitsS,file = 
> "test", times = 10000)
> R> test$chr<-as.character(c("chr1","chr2"))
> R> score2wig(test, file = "test.wig", times = 10000)
> 
> Let me know if you have any problem.
> Jose
> 
> 
> 
> -----Mensaje original-----
> De: bioc-sig-sequencing-bounces at r-project.org en nombre de Chen-Yi 
> ChenEnviado el: vie 26/03/2010 23:02
> Para: bioc-sig-sequencing at r-project.org
> Asunto: [Bioc-sig-seq] ChIP-seq analysis in normalization/peak 
> calling between sample and control
> 
> Hi all,
> After going through all the ChIP-seq pipeline in ht-seq, I finally 
> come down to normalization/peak calling. Interestingly, ht-seq 
> seems to not have a standard normalization and peak calling 
> algorithm between sample and control. I've read through the 
> previous threads about "peaks calling," and people suggest all 
> different things.
> So I've tried the following packages:
> chipseq -> using the cutoff as island/peaks calling, and there is 
> nothing about normalization technique. From my understanding, I 
> thought we need a normalized data from sample and control in order 
> to do this, and I certainly don't know how to normalize them in ht-
> seq.SPP -> didn't work, it returned a NaN out of range error when 
> doing the enrichment (peak calling) calculation.
> CSAR -> didn't seem to be available in installation through 
> biocLite, but I downloaded and installed it manually. It didn't 
> seem to work well with ShortRead. (at least I have no idea how to 
> do it), and again, wig file output didn't visualize on UCSC genome 
> browser.ChIPseqR -> I didn't go in to it that much, but it seemed 
> to be the "simulating package"
> ChIPsim -> similar story as ChIPseqR
> PICS -> I visited their website and they mentioned it should be 
> available through bioconductor website, but I didn't see any PICS 
> packages on bioconductor website.
> 
> Is there a standard (or simple) normalization technique/peak 
> calling package in ht-seq that we can use?
> I am not a statistician, so any suggestions on this 
> normalization/peak calling would really help our analysis.
> 
> Thanks a bunch!
> 
> -Charlie-
> 
> _______________________________________________
> Bioc-sig-sequencing mailing list
> Bioc-sig-sequencing at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioc-sig-sequencing
> 
> 
> 
>

_______________________________________________
Bioc-sig-sequencing mailing list
Bioc-sig-sequencing at r-project.org
https://stat.ethz.ch/mailman/listinfo/bioc-sig-sequencing