SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics



Similar Threads
Thread Thread Starter Forum Replies Last Post
mulitple/split/chimeric reads from bwa bwasw long read aligner and SNP calling Maulik23 Bioinformatics 1 08-25-2013 06:22 AM
SNP calling with paired and single read data mht Bioinformatics 7 10-25-2012 01:17 AM
PubMed: ngs_backbone: a pipeline for read cleaning, mapping and SNP calling using Nex Newsbot! Literature Watch 0 06-05-2011 06:00 AM
What cut off does GATK snp calling use for mapping quality score? foxyg Bioinformatics 0 09-25-2010 11:19 AM

Reply
 
Thread Tools
Old 05-08-2015, 07:04 AM   #1
lre1234
Senior Member
 
Location: US

Join Date: Aug 2011
Posts: 106
Default nonunique read mapping and snp calling

Hi all,
We are having a little debate here that we would like some other insights into.

The problem is that we are doing targeted sequencing of a few hundred genes, and plan on looking for snps within those genes. the issue is that a subset of the genes, have some pseudogenes, and/or repetitive regions within them, that would prevent the reads from mapping uniquely to those genes. So we have two choices:

first would be to allow reads to map to multiple locations in the genome, which would include those repetitive regions within our target genes, and then do SNP calling specifically within the target regions.
or
second, be conservative with the alignments, and allow for reads to only map uniquely in the genome, and then do the SNP calling.

From our discussions, the advantage of the first approach would be that there may be some useful snps within those regions that would be able to see, which may be important for our disease or phenotype. We seem to be on the fence as to which approach to do and which would be better.

Just wanted to know if anyone may have a thought about this.
lre1234 is offline   Reply With Quote
Old 05-08-2015, 08:38 AM   #2
Brian Bushnell
Super Moderator
 
Location: Walnut Creek, CA

Join Date: Jan 2014
Posts: 2,707
Default

I suggest throwing out all reads that do not map uniquely before doing variant-calling. If that excludes too much data, then you should probably use longer reads and a longer insert size for pairs.
Brian Bushnell is offline   Reply With Quote
Old 05-08-2015, 08:42 AM   #3
Richard Finney
Senior Member
 
Location: bethesda

Join Date: Feb 2009
Posts: 700
Default

How many samples?
Richard Finney is offline   Reply With Quote
Old 05-08-2015, 08:46 AM   #4
lre1234
Senior Member
 
Location: US

Join Date: Aug 2011
Posts: 106
Default

Quote:
I suggest throwing out all reads that do not map uniquely before doing variant-calling. If that excludes too much data, then you should probably use longer reads and a longer insert size for pairs.
we are using 2x150 PE reads and a 300bp insert size (I think, but need to double check). I guess, my question would be why throwout the reads? What would (in your opinion) be the issue with doing the snp calling in those regions?

Quote:
How many samples?
~1000 samples
lre1234 is offline   Reply With Quote
Old 05-08-2015, 08:56 AM   #5
fanli
Senior Member
 
Location: California

Join Date: Jul 2014
Posts: 198
Default

Let's say that some of the non-uniquely mapped reads actually come from the pseudogenes. Then you'd probably get false positive SNP calls in your target regions

Agree w/ BB, the standard is to throw out non-unique mappers.
fanli is offline   Reply With Quote
Reply

Tags
align, non-unique mapping, repeat element, unique alignments

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 04:19 PM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2021, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO