peak calling error

[tim]

Hi,

I am very interested in analyzing my one sample chip-seq data using
this method. I use the "process_readfiles.pl" to generate the bin
level data from sam input and do the analysis in chromosomewise. It
works fine for chr7 and not for chr1 (I just tried these two chr
now).

(1) OneSampleBinData_chr1 <- readBins(type = c("chip","M","GC","N"),
fileName = c("./chipseq_data_bin/S1/
chr1_S1_Mapped.sam_fragL250_bin50.txt","./hg18/
chr1_map_fragL200_bin200.txt","./hg18/chr1_GC_fragL200_bin200.txt","./
hg18/chr1_N_fragL200_bin200.txt"))

Warning messages:
1: In Y > 0 & M == 0 :
longer object length is not a multiple of shorter object length
2: In Y > 0 & M == 0 :
longer object length is not a multiple of shorter object length

(2) OneSampleFit_chr1 <- mosaicsFit(OneSampleBinData_chr1,
analysisType = "OS") ## neither error nor warning for this step

(3) OneSamplePeak_chr1 <- mosaicsPeak(OneSampleFit_chr1,
signalModel="1S", FDR = 0.05, maxgap = 250, minsize = 50, thres = 10)
Info: use one-signal-component model.
Info: calculating posterior probabilities...
Error in 0:Ymax : NA/NaN argument

Thanks for your help,
tim

[Dongjun Chung]

Hi Tim,

Which mosaics version are you using? I updated mosaics package (ver.
1.2.5) and it is now on bioconductor:

http://www.bioconductor.org/packages/release/bioc/html/mosaics.html

I think that some of the errors you mentioned should be fixed in the
updated version. Can you try again with the new version and let me
know how it works? There are many new features and bug fixes so I
highly encourage you to use the new version. The some of new features
are:

- process aligned read files within R environment (constructBins()
function).
- genome-wide analysis (don't need to analyze each chromosome one by
one).
- much faster calculation using internal C++ implementation using
Rcpp.
- supports parallel processing using multicore.

I believe that if you do genome-wide analysis using the new version,
many of your problems will be gone.

If you are using R 2.14.x, the new version will be installed by
default with the commands:

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

If you are using R 2.13.x, older version of mosaics (ver. 1.0.1) will
be installed by default. In this case please just download mosaics
package from the webpage & install it.

Dongjun

[Yingtao Bi]

Hi Dongjun,
 
 If I want to do genome-wise analysis, do I need to have mappability, GC content, ambiguity (N) files in genome wide? If yes, how to generate them? If No, how to incorproate those files in readBins? If the binsize used to generated to the bin-level data is 50, could I use mappability, GC content, ambiguity (N) files generated with binsize equal to 200?

Thanks,
Tim

[Dongjun Chung]

Hi Tim,

Sorry for the late response.

Yes, that's correct. You need to have genome-wide versions of
mappability, GC content, and sequence ambiguity score files. At this
point, we only uploaded genome-wise files to our website. Can you let
me know which bin size you want to use? Files need to be re-generated
for different bin size. And I guess that you are working on hg18. If
you confirm these, I'll generate corresponding files.

Thanks,
Dongjun

On Mar 15, 1:53 pm, Yingtao Bi <ybi.wis...@gmail.com> wrote:
> Hi Dongjun,
>
>  If I want to do genome-wise analysis, do I need to have

> mappability<http://www.stat.wisc.edu/%7Ekeles/Software/hg18_map_unireads.tar.gz>,
> GC content <http://www.stat.wisc.edu/%7Ekeles/Software/hg18_GC.tar.gz>,
> ambiguity
> (N) <http://www.stat.wisc.edu/%7Ekeles/Software/hg18_N.tar.gz> files in

> genome wide? If yes, how to generate them? If No, how to incorproate those
> files in readBins? If the binsize used to generated to the bin-level data
> is 50, could I use

> mappability<http://www.stat.wisc.edu/%7Ekeles/Software/hg18_map_unireads.tar.gz>,
> GC content <http://www.stat.wisc.edu/%7Ekeles/Software/hg18_GC.tar.gz>,
> ambiguity
> (N) <http://www.stat.wisc.edu/%7Ekeles/Software/hg18_N.tar.gz>
> filesgenerated with binsize equal to 200?
>
> Thanks,
> Tim