Newbler's algorithm -> overlap or de Brujin graph?

[CImbusch]

Hi all,

for my diploma thesis I wanted to roughly describe how Newbler works. Reading the paper from Ewan Birney, he explains the de Brujin graph structure and says:

"The first assembler to exploit this technology was Roche's 454 assembler, Newbler, which adapted the scheme specifically to handle the main source of error in 454 sequencing—namely, ambiguity in the length of homopolymer runs."

Reading the paper from Margulies et al. from 2005 and its supplementary is mentioning an 'overlapper' module which performs an all-against-all fragment comparison. Sounds to me it uses an overlap graph.

Now I'm very confused about what it is actually using. Can somebody help me out?

Cheers,
Charles

[francesco.vezzi]

HI
Newbler is definitely base on the Overlap Layout Consensus methodology. I suggest to you to read the survey "Assembly algorithms for next-generation sequencing data." by Miller.

Newbler works with reads of length greater than 250. A De Bruijn Graph approach is feasible but when reads are so long it is better use an overlap layout method in order to avoid a great number of false positives...

[CImbusch]

Hi Francesco,

thanks for the reply. The article "Assembly algorithms for next-generation sequencing data." by Miller made me first suspecious. Yeah, Newbler seems to be an OLC assembler, but I'm still confused what Ewan Birney was meaning by his statement.

Cheerio,
Charles

[CImbusch]

Who ever is interested how Newbler works, I found a nice blog:

http://contig.wordpress.com/

Cheers,
Charles