Question

SAAP-BS data analysis (RRBS) - BiSeq use and Coverage Histogram

0

Entering edit mode

maria.maqueda • 0

@mariamaqueda-7673

Last seen 9.5 years ago

European Union

Hello all,

I am struggling with the DMR analysis of some RRBS data already processed with SAAP-BS program. I do not have the SAM/BAM files, just a merged csv file with the methylC, totalC and Ratio per sample and CpG site (around 2.5M).

I am new to this type of analysis (previously worked with expression analysis) and I have decided to use BiSeq for obtaining DMR. My first question is: I have to create by hand the BSraw object, is "totalReads" equivalent to "totalC" and "methReads" equivalent to "methC"?

On the other hand I have supposed that "totalC" from SAAP-BS report is equivalent to the coverage mentioned in RRBS/WGBS publications. Assuming my supposition is right (if not, please correct me!!!), I have found there are very high coverage values (up to maximum value of 16k - this located in chromosome 21, around 4k CpG sites with coverage > 200). Is this to be expected?

If you could bring some light here I would really appreciate. Many thanks in advance!

Maria

biseq RRBS SAAP-BS coverage • 2.3k views

ADD COMMENT • link 9.6 years ago • updated 9.5 years ago maria.maqueda • 0

score 0 · Answer 1 · 2015-04-28

Hi Maria, I don't know SAAP-BS, but from the column names I suppose that you are right: totalC corresponds to totalReads and methC and to methReads. CpG sites with very high coverages are to be expected. At least I have never seen an RRBS data set without it. In the BiSeq pipeline it is suggested to limit the coverage, e.g., to the 0.9 quantile (page 15): http://bioconductor.org/packages/release/bioc/vignettes/BiSeq/inst/doc/BiSeq.pdf Cheers, Katja ----- Original Message ----- > From: "maria.maqueda [bioc]" <noreply@bioconductor.org> > To: katjah@stanford.edu > Sent: Tuesday, April 28, 2015 8:53:15 AM > Subject: [bioc] SAAP-BS data analysis (RRBS) - BiSeq use and Coverage > Histogram > Activity on a post you are following on support.bioconductor.org > User maria.maqueda wrote Question: SAAP-BS data analysis (RRBS) - BiSeq use > and Coverage Histogram : > Hello all, > I am struggling with the DMR analysis of some RRBS data already processed > with SAAP-BS program. I do not have the SAM/BAM files, just a merged csv > file with the methylC, totalC and Ratio per sample and CpG site (around > 2.5M). > I am new to this type of analysis (previously worked with expression > analysis) and I have decided to use BiSeq for obtaining DMR. My first > question is: I have to create by hand the BSraw object, is "totalReads" > equivalent to "totalC" and "methReads" equivalent to "methC"? > On the other hand I have supposed that "totalC" from SAAP-BS report is > equivalent to the coverage mentioned in RRBS/WGBS publications. Assuming my > supposition is right (if not, please correct me!!!), I have found there are > very high coverage values (up to maximum value of 16k - this located in > chromosome 21, around 4k CpG sites with coverage > 200). Is this to be > expected? > If you could bring some light here I would really appreciate. Many thanks in > advance! > Maria > You may reply via email or visit SAAP-BS data analysis (RRBS) - BiSeq use and Coverage Histogram

score 0 · Answer 2 · 2015-05-07

0

Entering edit mode

maria.maqueda • 0

@mariamaqueda-7673

Last seen 9.5 years ago

European Union

Hi again Katja,

Following your suggestion to limit the coverage as indicated in the BiSeq vignette, I obtained the following error message while executing limitCov function:

"Error in methReads(object)[indCov] <- as.integer(round(fraction * maxCov)) :
NAs are not allowed in subscripted assignments"

My "clust.unlim" object contains missing values (percentage of NAs ranges between 0.13% and 7.7% which I do not really know if this is quite normal or not for a sample size of 20.

I have tried to smooth the data without limiting the coverage and have seen that this "NAs issue" gets worse. Could this be due to the meth levels prediction? Shall I remove this NA readings prior to constructing the CpG clusters?

Many thanks in advance for any advice!!

Maria

ADD COMMENT • link 9.5 years ago maria.maqueda • 0

0

Entering edit mode

Hi Maria, It sounds like you have NAs in your CpGs that are not covered. Please replace the NAs by a 0 in the totalReads and methReads slots. Does that help? Cheers, Katja ----- Original Message ----- > From: "maria.maqueda [bioc]" <noreply@bioconductor.org> > To: katjah@stanford.edu > Sent: Thursday, May 7, 2015 3:56:08 AM > Subject: [bioc] A: SAAP-BS data analysis (RRBS) - BiSeq use and Coverage > Histogram > Activity on a post you are following on support.bioconductor.org > User maria.maqueda wrote Answer: SAAP-BS data analysis (RRBS) - BiSeq use and > Coverage Histogram : > Hi again Katja, > Following your suggestion to limit the coverage as indicated in the BiSeq > vignette, I obtained the following error message while executing limitCov > function: > > "Error in methReads(object)[indCov] <- as.integer(round(fraction * maxCov)) : > NAs are not allowed in subscripted assignments" > My "clust.unlim" object contains missing values (percentage of NAs ranges > between 0.13% and 7.7% which I do not really know if this is quite normal or > not for a sample size of 20. > I have tried to smooth the data without limiting the coverage and have seen > that this "NAs issue" gets worse. Could this be due to the meth levels > prediction? Shall I remove this NA readings prior to constructing the CpG > clusters? > Many thanks in advance for any advice!! > Maria > You may reply via email or visit > A: SAAP-BS data analysis (RRBS) - BiSeq use and Coverage Histogram

ADD REPLY • link 9.5 years ago Katja Hebestreit ▴ 130

0

Entering edit mode

Hi Katja,

Yes, this helps! Many thanks for answering so quick! I am just smoothing the meth data. Let's see what I have!

Thanks again,

Cheers

Maria

ADD REPLY • link 9.5 years ago maria.maqueda • 0