On 18/11/13 06:16 PM, Bastien Chevreux wrote:
> On 18 Nov 2013, at 19:47 , Sébastien Boisvert <[email protected]> 
> wrote:
>> I (probably) found the issue.
>> […]
>> Working on a patch now.

Hi Bastien,

(I CC'ed the list)

Good to hear from the author of MIRA !


>
> Thanks for that.
>
> Off-github: from all the pure nextgen (Illumina, late 454 reads) assemblers I 
> tried, I like Ray best so far. Good work there. It does have its fair share 
> of misassemblies
>when I run it on my horror projects I have in the archive, but definitely less 
>than a couple of other assemblers I tested. Though it completely fails with 
>early 454
>and current gen Ion data, but then again dbg assemblers are not really made to 
>cope with pesky homopolymers and carry forward errors.

Yes. Ray lacks some sort of bubble pass-through for these,

>
> I noticed Ray had also quite some trouble with the uneven coverage as 
> produced by Nextera kit preps, but I think this is inevitable for assemblers
>which somehow try to deduce repetitiveness from (k-mer or other) coverage.
>

I agree the process could be refined.

For uneven coverage: https://github.com/sebhtml/ray/issues/132

Last year, I started to work on an interactive visualizer to look at these sort 
of things.


For instance, you can take a look at uneven coverage at 
http://genome.ulaval.ca:10208/client/?map=0&section=0&region=381&location=304&depth=10&zoom=0.6154458722734157
by pressing the "Graphs" button.

The 2.x.x series is way better with uneven coverage and this is one of the 
changes we made to tackle
metagenomes. see http://genomebiology.com/2012/13/12/R122

> Ray will be a good validation assembler for me … and for projects where I’ll 
> have to ditch MIRA due to the number of reads / size of the genome :-)
>
> Best,
>    Bastien
>

For mini-ranks:

Basically, I have been playing around with various approaches to program 
scalable tools.
Some of these approaches are error-prone while others are easier to master. So 
far, I can say
that RayPlatform (some sort of framework on top of Ray) implements 3 models:

- slave modes / master modes / message tags (RayPlatform Plugin API)
- mini-rank runtime (compatible with RayPlatform Plugin API)
- actor model (RayPlatform Actor Playground)

I recently implemented a tool to compare many samples without a reference using 
kmers
and the de Bruijn graph with this actor model. The code is much easier to read 
than distributed
code that does not use this actor model.


I believe you are located in Europe.
will be presenting Ray and Ray Cloud Browser at Beatles and Bioinformatics in 
Liverpool, UK.
on 27 November 2013.
http://pathogenomics.bham.ac.uk/blog/2013/10/beatles-and-bioinformatics/


------------------------------------------------------------------------------
Shape the Mobile Experience: Free Subscription
Software experts and developers: Be at the forefront of tech innovation.
Intel(R) Software Adrenaline delivers strategic insight and game-changing 
conversations that shape the rapidly evolving mobile landscape. Sign up now. 
http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk
_______________________________________________
Denovoassembler-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/denovoassembler-users

Reply via email to