Sam flags for bwa-aligned paired end reads with identical + / - strand coordinates

spark · 03-09-2011, 04:00 AM

I'm looking at some paired end WGS data from a library with very small insert sizes, meaning that many of the DNA fragments are sequenced fully in both directions.

Reads were aligned to a template genome using BWA. For many reads, the insert size is zero (i.e. forward and reverse strand coordinates are identical) These reads are being flagged by bwa sampe as improperly paired (sam flags 81 & 161 or 97 & 145), whereas I think they should count as properly paired.

My question is whether downstream tools (specifically GATK realignment and SNP calling) will exclude these reads, and if so what I would need to do to get round the problem. (One workaround solution would be to extract these improperly paired reads, trim an extra base from the 3' end, then realign.)

As an example:
HWUSI-EAS174:5:102:602:453#0 97 chr1 66904 15 22M = 66904 22 CCAGGAGGCAGCAGCAGTAGCC B@?AA=A@@=A><>=A94>==9
HWUSI-EAS174:5:102:602:453#0 145 chr1 66904 15 22M = 66904 -22 CCAGGAGGCAGCAGCAGTAGCC B=6B4@A?69@9<B?;7=BABA

Thanks,

Stephen

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Paired-end Bam from single-end aligned sam	ramouz87	Bioinformatics	4	08-17-2011 12:55 PM
How to infer Illumina paired-end strand specificity from SAM output?	David Harmin	Bioinformatics	0	02-16-2011 08:34 AM
BWA alignment for paired end reads	AvinashP	Genomic Resequencing	2	06-08-2010 03:11 AM
TopHat SAM - Expressing Paired End Multi-reads	Bio.X2Y	Bioinformatics	2	05-28-2010 08:07 AM
losing %reads aligned with Bowtie paired end analysis	Batool	Bioinformatics	0	04-21-2010 09:14 AM

03-09-2011, 04:00 AM	#1
spark Junior Member Location: Dublin Join Date: Mar 2009 Posts: 6	bwa bug? Paired end reads with insert size of 0 flagged as improperly paired I'm looking at some paired end WGS data from a library with very small insert sizes, meaning that many of the DNA fragments are sequenced fully in both directions. Reads were aligned to a template genome using BWA. For many reads, the insert size is zero (i.e. forward and reverse strand coordinates are identical) These reads are being flagged by bwa sampe as improperly paired (sam flags 81 & 161 or 97 & 145), whereas I think they should count as properly paired. My question is whether downstream tools (specifically GATK realignment and SNP calling) will exclude these reads, and if so what I would need to do to get round the problem. (One workaround solution would be to extract these improperly paired reads, trim an extra base from the 3' end, then realign.) As an example: HWUSI-EAS174:5:102:602:453#0 97 chr1 66904 15 22M = 66904 22 CCAGGAGGCAGCAGCAGTAGCC B@?AA=A@@=A><>=A94>==9 HWUSI-EAS174:5:102:602:453#0 145 chr1 66904 15 22M = 66904 -22 CCAGGAGGCAGCAGCAGTAGCC B=6B4@A?69@9<B?;7=BABA Thanks, Stephen Last edited by spark; 03-09-2011 at 12:55 PM. Reason: Changing title to something clearer