Knon SNPs, non-directional, paired vs. single end reads,

[Zhiyuan Chen]

Dear Dr. Liu,

Thank you for developing the fantastic tool and spending time for answering questions! I have several questions for you related to BisSNP. Forgive me if these are stupid questions since I'm new to the field.

I'm analyzing the global DNA methylation of bovine fetal muscle. The library is non-directional and sequenced on a Illumina Hiseq 2000 with 2X100bp paired-end reads. The average coverage of initial sequencing is about 10X coverage (1 lane, ~150 million read pairs). I'm using BisSNP to correct the methylation level from Bismark and call SNPs for allele-specific DNA methylation.

Question 1):

         Known SNPs are required in the recalibration step and strongly recommended in the genotype step. However, unlike human, there is no available SNP database for the cow breeds we used in our study. We only have about 100 high confident SNPs of our animals which were previously identified by Sanger sequencing. How inaccurate it will be for methylation and SNP call if we just skip the recalibration step? If the 100 known SNPs provided, it wouldn't make a difference for those SNPs beyond the 100, right?

Question 2):
 
         We used non-directional protocol. I noticed that non-directional option has been added to the BisSNP, although it's not included in the manual. Adding -non directional option should not change any interpretation of the filtered.cpg.vcf and filter.snp.vcf output, right?

Question 3):

         When I trimmed paired end reads (read1 and read2), if read1 is of very low quality (e.g. replaced by N after trimming), read2 was aligned separately as a single end read. The reason why I do this is because bismark will not align one read if another read in the pair is of very low quality.
         Therefore, I ended up with two bam file: one is the paired-end alignment (both reads in a pair are of good quality), the other is single-end alignment (one read in a pair is good quality, the other is bad).
          Can I merge the two bam files into a single one for BisSNP pipeline assuming it's a signle-end alignment file?

Thank you so much for reading this!
Looking forward for your comments!

Happy Labor Day!
Thanks!

Zhiyuan

[ping]

Dear Zhiyuan,

Thanks for your interests in our software! The primary purpose of Bis-SNP was developed for directional protocol WGBS. Later on, i adapt some code for non-directional protocol, but i have never test the SNP calling accuracy by using SNP array for non-directional protocol. So I am not sure how well it will performs for your library.

Q1: known SNP in recalibration step is used to avoid to include position in known SNP as a error rate, the effect really depends on your organism. For mouse, it will affect a lot on recalibration step since high SNP rate in most of mouse strain.

Q2: should be. But I am not sure the accuracy. It will be better to check it through IGV browser. or if you have some SNP array on the same sample.

Q3: For your second single-end alignment library, if your reads was marked as single end, it does not affect at al. But if your reads was marked as paired-end, and 1st end was unmapped, Then you have to enable the option:  -USE_BADLY_MATED_READS.

let me know if you have further questions.

Best,

yaping

--
You received this message because you are subscribed to the Google Groups "bissnp-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bissnp-help...@googlegroups.com.
To post to this group, send email to bissn...@googlegroups.com.
Visit this group at http://groups.google.com/group/bissnp-help.
For more options, visit https://groups.google.com/d/optout.

--

Yaping Liu Ph.D.

Postdoctoral Associate

Manolis Kellis Lab

Computer Science and Artificial Intelligence Lab (CSAIL)
Massachusetts Institute of Technology

Broad Institute of MIT and Harvard

Office Phone: +1 617 324 8436 
Email: 

lypin...@gmail.com

yap...@mit.edu

lyp...@broadinstitute.org

[Zhiyuan Chen]

Thank you for your quick and clear reply. It's a lot of help.

Related to your answer about marked paired-end reads in BAM file, how to determine if it's marked or not?

I compared the paired and single end alignment BAM, the only difference is the fields "RNEXT", "PNEXT", and "TLEN".

Here is an example:

                    Paired                                   Single
RNEXT           "="                                        "*"
PNEXT           "153893053"                        "0"
TLEN              "199" or "-199"                     "0"

Now I know how they are related to paired or single read, but did you use these fields to determine a read belongs to paired-end or single-end alignment?

Thanks!
Zhiyuan

[ping]

I use the flag in the second column as described in SAM format here:

http://samtools.github.io/hts-specs/SAMv1.pdf

you can also check how these flag number related with reads status here:

http://picard.sourceforge.net/explain-flags.html

yaping

--

You received this message because you are subscribed to the Google Groups "bissnp-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bissnp-help...@googlegroups.com.
To post to this group, send email to bissn...@googlegroups.com.
Visit this group at http://groups.google.com/group/bissnp-help.
For more options, visit https://groups.google.com/d/optout.

[Zhiyuan Chen]

  Yes!!! The single end read alignment is not labeled as paired end. Thank you very much!

[Raphaël Poujol]

I am sorry because I feel I already ask that but there is no way I can not retrieve the option for non directional library.
Could you remid me how to launch BisSNP to call methylation on non directionnal protocol ?

Thank you !

[ping]

Hi Raphael,

Sorry for the late reply! The option for non-directional is "-nonDirectional"

You can retrieve all of the options by: Java -jar Bis-SNP.jar -T BisulfiteGenotyper

without any other options.

Sorry for the inconvenience, i should update the manual when i have time...

yaping

Yaping Liu, Ph.D.