Input files for the Celera Assembler?

[dan]

@Dan,

my suggestion was to either use the bambus mate pair output or a TraceArchive formatted XML file for use with "toAmos", like

$ toAmos -s x.fasta -q x.qual -m test.mates -o test.a.afg

or

$ toAmos -s x.fasta -q x.qual -x test.xml -o test.a.afg

but, of course, you have to create these files before.

It is probably the easiest way to write your own "your_data2TraceArchive.pl" and then convert it to whatever you want, e.g. CA format using "tracedb-to-frg.pl".

Yeah, that's what I'm struggling with... Any reference for the bambus format on the off chance that its simpler? I can't help thinking there is a canned solution in the AMOS docs if only I could find it.

[new300]

Celera Assembler has been optimised for at least 454FLX length reads; I have used it for shorter GS20 as well, but it may fail. All these assemblies were hybrid assemblies, not 454-only ones.

That's interesting, I'm surprised it copes with the elevated indel/homopolymer run error rate.

[sklages]

@Dan,

my suggestion was to either use the bambus mate pair output or a TraceArchive formatted XML file for use with "toAmos", like

$ toAmos -s x.fasta -q x.qual -m test.mates -o test.a.afg

or

$ toAmos -s x.fasta -q x.qual -x test.xml -o test.a.afg

but, of course, you have to create these files before.

It is probably the easiest way to write your own "your_data2TraceArchive.pl" and then convert it to whatever you want, e.g. CA format using "tracedb-to-frg.pl".

@new300,

Celera Assembler has been optimised for at least 454FLX length reads; I have used it for shorter GS20 as well, but it may fail. All these assemblies were hybrid assemblies, not 454-only ones.

hth,
Sven

[new300]

Are you planning to use the Celera assembler with short reads? Because I don't think it would work very well.

[dan]

Hi,

I have sequence reads as a fasta file with a fasta quality file. I am converting these into .frg input for the Celera Assembler (CA) using AMOS,

toAmos -s test.fasta -q test.qual -o test.afg

and then:

amos2frg -i test.afg

However, the above strategy does not recognize the 'mate pairs' in my data (paired end reads) that are 'linked' using the St Louis naming convention (basically read0012.b is pared with read0012.g, the .b and the .g denoting the forward and the backward read, respectively).

On the AMOS mailing list I got a reply to a similar question from Sven Klages, which told me that CA is usually passed a separate "linkage information" file in addition to the .frg input created above. His suggestion was to create linkage data in 'Trace Archive' format*, and then convert that to input for CA using AMOS again.

Has anyone tackled this problem before? Is there any code 'off the shelf' to create the mate pair data?


* http://www.ncbi.nlm.nih.gov/Traces/t...#header-global


Cheers,
Dan.


P.S. Any other good mailing lists for sequence assembly?

If you know any, I'll add them here:

http://biodatabase.org/index.php/List_of_mailing_lists_for_biologists