SEQanswers

Go Back   SEQanswers > Bioinformatics > Bioinformatics

Similar Threads
Thread Thread Starter Forum Replies Last Post
samtools flagstat output nguyendofx Bioinformatics 23 01-22-2014 05:45 AM
samtools flagstat output manore Bioinformatics 11 07-24-2013 12:19 AM
SAMtools flagstat output interpretation a2z@blr Bioinformatics 2 10-20-2011 02:23 PM
Samtools flagstat Anelda Bioinformatics 0 09-26-2011 04:55 AM
samtools flagstat bair Bioinformatics 3 05-28-2010 07:15 AM

Reply
 
Thread Tools
Old 11-24-2011, 07:22 AM   #1
Orr Shomroni
Member
 
Location: Netherlands

Join Date: Oct 2011
Posts: 26
Question Samtools flagstat - no duplicates?

I'm having some problems with samtools flagstat (which surprisingly should be straightforward as it has no parameters that can be wrongly defined). Samtools flagstat is used to return summaries of the flags produced within bam files. I was given intensities from exome sequencing, converted them into fastq files and removed reads with low quality (fastq format). I ran an alignment on fastq files using BWA, and got the following weird results from flagstat:

109638024 + 0 in total (QC-passed reads + QC-failed reads)
0 + 0 duplicates
108394681 + 0 mapped (98.87%:nan%)
109638024 + 0 paired in sequencing
54819012 + 0 read1
54819012 + 0 read2
106794832 + 0 properly paired (97.41%:nan%)
108037629 + 0 with itself and mate mapped
357052 + 0 singletons (0.33%:nan%)
1071345 + 0 with mate mapped to a different chr
977974 + 0 with mate mapped to a different chr (mapQ>=5)

Common biological knowledge dictates that every sequencing should produce 10-15% PCR or optical duplicates, which is why I was surprised when I got 0 duplicates (I tried it on several samples in the same run, and they all gave 0 + 0 duplicates, which makes me feel something is wrong with flagstat). It got even weirder after I did samtools rmdup and retrieved the flagstat for that bam file:

59158619 + 0 in total (QC-passed reads + QC-failed reads)
0 + 0 duplicates
57915276 + 0 mapped (97.90%:nan%)
59158619 + 0 paired in sequencing
26999098 + 0 read1
32159521 + 0 read2
56923778 + 0 properly paired (96.22%:nan%)
57745533 + 0 with itself and mate mapped
169743 + 0 singletons (0.29%:nan%)
727165 + 0 with mate mapped to a different chr
672895 + 0 with mate mapped to a different chr (mapQ>=5)

So about half of the original reads are kept, i.e. there are some duplicates, but they were not flagged in bwa. How could this have happened?
__________________
"Though it may seem that all's been said and done, originality still lives on" - some unoriginal guy who had nothing better to write as his signature
Orr Shomroni is offline   Reply With Quote
Old 11-24-2011, 08:36 AM   #2
cjp
Member
 
Location: Cambridge, United Kingdom

Join Date: Jun 2011
Posts: 58
Default

You'll only get duplicates in the BAM file if they are marked as duplicates. BWA doesn't mark them and samtools rmdup removes them. Try using picard MarkDuplicates

java -jar ${picard_dir}/MarkDuplicates.jar I=${name}_bwa.bam O=${name}_bwa_dedup.bam M=${name}_bwa_dedup_metrics.txt

Chris
cjp is offline   Reply With Quote
Old 11-24-2011, 12:59 PM   #3
Orr Shomroni
Member
 
Location: Netherlands

Join Date: Oct 2011
Posts: 26
Default

Well, I have considered a dodgy trick: if I subtract the number of reads in the bam rmdup file from the number of reads in the original bam file, I should get back the number of duplicates removed, right?

One thing I'm still not certain about: BWA aligns the reads to get a bam file, and it doesn't flag any duplicates. Does rmdup than find those duplicates without them being previously marked/flagged? If so, how is this done?
__________________
"Though it may seem that all's been said and done, originality still lives on" - some unoriginal guy who had nothing better to write as his signature
Orr Shomroni is offline   Reply With Quote
Old 11-25-2011, 01:46 AM   #4
cjp
Member
 
Location: Cambridge, United Kingdom

Join Date: Jun 2011
Posts: 58
Default

It looks for reads that have the same start position and then keeps the best one based on sequence quality scores - it's a fairly basic approach. Also look at this thread which recommends to use picard:

http://seqanswers.com/forums/showthread.php?t=5424

BWA is kept light weight and doesn't do things like marking duplicates or even making BAM files as some applications don't need to mark duplicates - e.g., in ChIP-seq.

Chris
cjp is offline   Reply With Quote
Reply

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 02:44 PM.


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