Re: [Bio-bwa-help] color-space support to be dropped in version 0.6

[Brad G. <bg...@cs...>]

Hi Heng,

I have some experience with color-space alignment
of a complete human genome. For my application
we had to allow at least 2 nucleotide SNPS for any
short read. This required allowing for 4 color-space
polymorphisms, which made the alignment very
slow. Not quite prohibitive, but not really competitive.

We used BWA because it seemed the only fast aligner
at the time that:
1) could accept read data from any of the several technologies
     we had to compare (short and long reads, Sanger, Illumina, 454,
     SOLiD).
2) Could handle a whole-genome human deep sequences efficently.
3) Was in common used at the time (V5.5-5.7 ish).
4) Provided both read quality and alignment quality
     phred scores.

I think it would be helpful to maintain colorspace support
in BWA for this reason (comparison of multiple sequencing
technologies), but performance really needs to be improved
for color reads. If we had to do it again today, we might well
go with BFAST.

--Brad



On 11/21/2011 11:41 AM, Brent Pedersen wrote:
> On Sat, Nov 19, 2011 at 12:28 PM, Heng Li<lh...@sa...>  wrote:
>> The color-space alignment is not working in 0.6.0. Perhaps it is not so hard to make it work again, but bwa may not work well with solid reads all the time. Actually I have never evaluated this myself. 0.5.10 should work solid data.
>>
>> Any objections? Do you think it is worth keeping the color-space support in bwa?
>>
>> Thanks,
>>
>> Heng
> For what it's worth, I have tried a few aligners for colorspace on
> some targetted re-sequencing reads where I can gauge accuracy somewhat
> by % of reads on target.
> Most of the aligners have similar accuracy, given some parameter
> tuning, but BWA does not appear at the top in terms of number of reads
> mapped--at least for the parameters that I tried. does not output the
> CS, CQ tags so other options -- namely
> BFAST, shrimp and novoalignCS are more appealing.
> It also seems that the hash-based aligners are generally a bit better
> for colorspace.
>
> So, maybe better to focus efforts on base-space in BWA.
>
> my 0.02.
> -Brent
>
>
>> --
>>   The Wellcome Trust Sanger Institute is operated by Genome Research
>>   Limited, a charity registered in England with number 1021457 and a
>>   company registered in England with number 2742969, whose registered
>>   office is 215 Euston Road, London, NW1 2BE.
>>
>> ------------------------------------------------------------------------------
>> All the data continuously generated in your IT infrastructure
>> contains a definitive record of customers, application performance,
>> security threats, fraudulent activity, and more. Splunk takes this
>> data and makes sense of it. IT sense. And common sense.
>> http://p.sf.net/sfu/splunk-novd2d
>> _______________________________________________
>> Bio-bwa-help mailing list
>> Bio...@li...
>> https://lists.sourceforge.net/lists/listinfo/bio-bwa-help
>>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure
> contains a definitive record of customers, application performance,
> security threats, fraudulent activity, and more. Splunk takes this
> data and makes sense of it. IT sense. And common sense.
> http://p.sf.net/sfu/splunk-novd2d
> _______________________________________________
> Bio-bwa-help mailing list
> Bio...@li...
> https://lists.sourceforge.net/lists/listinfo/bio-bwa-help