Re: [Bio-bwa-help] BWA - relaxing parameters (minimum length fraction?)

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 454-5900

Hi Audra,

Heng can speak to this better than I can, but I don't think there's going
to be a way to specify that bwa mem only give you hits where 50% of the
read length maps to the reference (and, how would you define that for
paired-end alignments?). The scoring scheme is meant to reflect
"closeness", in a way that's not necessarily linearly related to coverage
of the hits on the reads, if that makes sense. I think your best bet is to
leave the scoring scheme alone, and write a script to parse the CIGAR
strings for each alignment (field 6, in the SAM format). Take a look at the
bottom of this <http://bio-bwa.sourceforge.net/bwa.shtml> page for
starters, and then this <http://samtools.github.io/hts-specs/SAMv1.pdf> for
the SAM specification, which will describe CIGAR strings. If you use R,
maybe look at Rsamtools, or pysam if you use Python ...

Hope that helps,

~Joe

On Fri, May 2, 2014 at 11:19 AM, Loy, Audra L <aud...@ma...>wrote:

> Hi, Heng,
>
> Sorry to bother you again, but I am now trying to use BWA mem as you
> instructed. However, I still cannot figure out how to relax the parameters
> as I would want them. Basically, if only 50% of the read maps to the
> reference transcript, I still want to count it. Does this make sense?
> Thanks!
>
> Audra
> ________________________________________
> From: Heng Li <lh...@me...>
> Sent: Friday, May 2, 2014 11:26 AM
> To: Loy, Audra L
> Cc: bio-bwa-help
> Subject: Re: [Bio-bwa-help] BWA - relaxing parameters (minimum length
> fraction?)
>
> I was not accurate: bwa-aln does not give you local hits with single-end
> reads. It may find local hits with mate rescue, but this is usually not
> sensitive enough when local hits are frequent.
>
> Use bwa-mem.
>
> Heng
>
> On May 2, 2014, at 12:23 PM, Loy, Audra L <aud...@ma...> wrote:
>
> > I was able to get local hits from bwa aln. These hits were actually
> quite similar to what I obtained in CLC. I'm confused….
> >
> > Audra
> > ________________________________________
> > From: Heng Li <lh...@me...>
> > Sent: Friday, May 2, 2014 10:06 AM
> > To: Loy, Audra L
> > Cc: Joseph Fass; bio-bwa-help
> > Subject: Re: [Bio-bwa-help] BWA - relaxing parameters (minimum length
> fraction?)
> >
> > Please try “bwa mem”, not “bwa aln”. Bwa-aln doesn't give you local hits.
> >
> > Heng
> >
> > On May 2, 2014, at 11:01 AM, Loy, Audra L <aud...@ma...>
> wrote:
> >
> >> Hi, Joe,
> >>
> >> Sorry about the lack of info. and thanks for the response. I am doing
> RNA-seq analysis with a reference transcriptome. I am using BWA aln and the
> only parameters I have changed are n =2 (max edit distance) and t = 20
> (threads). I left everything else at the default settings. What I am trying
> to figure out is what parameter determines how much of the read has to map
> before it is counted. For the purpose of my experiment, we want to try
> counting all reads that map at least 50%. I used to use CLC for mapping and
> would set minimum length fraction to 0.5. I was wondering what parameter in
> BWA might be similar. Like I said, I have read the documentation, but am
> struggling a bit with understanding the meaning behind each parameter.
> >>
> >> Again, I really appreciate any input. Thanks so much!
> >>
> >> Best,
> >> Audra
> >> From: Joseph Fass <jos...@gm...>
> >> Sent: Friday, May 2, 2014 9:45 AM
> >> To: Loy, Audra L
> >> Cc: BWA help
> >> Subject: Re: [Bio-bwa-help] BWA - relaxing parameters (minimum length
> fraction?)
> >>
> >> Hi Audra,
> >> It would probably help to show us the commands you've been using, so we
> can offer specific advice. And, what kind of situation are you looking to
> address? - Aligning to a close or distant relative's reference sequence? Is
> it a partial genome? Transcriptome? What do you mean by "count"? The more
> specific you can be, the better..
> >> ~Joe
> >> On May 1, 2014 4:21 PM, "Loy, Audra L" <aud...@ma...> wrote:
> >> Hi, all,
> >>
> >> So over the last few days, I have been teaching myself to use BWA. I am
> really getting the hang of it now, but am still trying to work out some of
> the parameters. I think things are too stringent at this point. One thing I
> would like to change is the minimum length fraction of the read that has to
> map to count. If this doesn't make sense, I basically want to count all
> reads that map 50% of their length to a particular transcript. I have read
> through the parameters in the manual, but cannot seem to figure out what to
> change to obtain this result. Any input would be greatly appreciated.
> >>
> >> Best,
> >> Audra
> >>
> >>
> ------------------------------------------------------------------------------
> >> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> >> Instantly run your Selenium tests across 300+ browser/OS combos.  Get
> >> unparalleled scalability from the best Selenium testing platform
> available.
> >> Simple to use. Nothing to install. Get started now for free."
> >> http://p.sf.net/sfu/SauceLabs
> >> _______________________________________________
> >> Bio-bwa-help mailing list
> >> Bio...@li...
> >> https://lists.sourceforge.net/lists/listinfo/bio-bwa-help
> >>
> >>
> ------------------------------------------------------------------------------
> >> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> >> Instantly run your Selenium tests across 300+ browser/OS combos.  Get
> >> unparalleled scalability from the best Selenium testing platform
> available.
> >> Simple to use. Nothing to install. Get started now for free."
> >>
> http://p.sf.net/sfu/SauceLabs_______________________________________________
> >> Bio-bwa-help mailing list
> >> Bio...@li...
> >> https://lists.sourceforge.net/lists/listinfo/bio-bwa-help
> >
>
>
> ------------------------------------------------------------------------------
> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
> Instantly run your Selenium tests across 300+ browser/OS combos.  Get
> unparalleled scalability from the best Selenium testing platform available.
> Simple to use. Nothing to install. Get started now for free."
> http://p.sf.net/sfu/SauceLabs
> _______________________________________________
> Bio-bwa-help mailing list
> Bio...@li...
> https://lists.sourceforge.net/lists/listinfo/bio-bwa-help
>

-- 
Joseph Fass
Lead Data Analyst
UC Davis Genome Center - Bioinformatics Core
http://bioinformatics.ucdavis.edu/
jn...@uc...
phone ~ 530.752.2698