Scaffolding output from a velvet assembly of SOLiD data

[xavier]

Hi,

I'm working on de novo assembly of a complex genome. I've been using velvet to generate the assembly of paired reads from SOLiD and now I need some software to generate the scaffolds.

I know Bambus, but it seems that I should parse the output from velvet to generate an input. I don't know how to do it, and maybe you know of some other software to achieve the scaffolding.

Regards,
Xavier.

[[email protected]]

Need Help

I know its too long to ask but i need help. Which software you use for scaffolding, was it Bambus? I am using Mosaik, a comparative assembler which generates .ace file (consensus). Can i use Bambus for scaffolding?

[xavier]

My first idea was to use Bambus for scaffolding, but at the moment I'm involved in other things and could not go deep with Bambus.

As soon as I have news about it I'll post it.

On the other hand we are working with other assemblers that seem to provide scaffolds in the final results, so if we get good results with them maybe we'll forget about velvet and Bambus.

Xavier.

[zhgai]

why not to use SOAPdenovo

[xavier]

For us at first it was not possible because SOAP developers only provide the binaries, and they didn't suit to our architecture. Recently they provided us with binaries for our architecture and currently we are testing it.

We have already ckecked the contigs generated by SOAP and it seems that we got excellent results. Now we are checking the scaffolds but we still don't have results.

On the other hand, it seems that abyss provides a tool to convert its results to ace format, so it may be used as an input for Bambus. We still couldn't try this, but that's the next step. If your assembler provides an ace file, then maybe you could use Bambus for scaffolding.

For velvet, we go on with the analysis just to compare the performance of each assembler, but it seems that ABYSS and SOAPdenovo generate much better results in a much lower time.

Best,
Xavier.

[[email protected]]

For us at first it was not possible because SOAP developers only provide the binaries, and they didn't suit to our architecture. Recently they provided us with binaries for our architecture and currently we are testing it.

We have already ckecked the contigs generated by SOAP and it seems that we got excellent results. Now we are checking the scaffolds but we still don't have results.

On the other hand, it seems that abyss provides a tool to convert its results to ace format, so it may be used as an input for Bambus. We still couldn't try this, but that's the next step. If your assembler provides an ace file, then maybe you could use Bambus for scaffolding.

For velvet, we go on with the analysis just to compare the performance of each assembler, but it seems that ABYSS and SOAPdenovo generate much better results in a much lower time.

Best,
Xavier.

But Bambus needs contig and mate information files as input. I have ace which i can convert to contig files "ace2contig" but how to get "mate file"?

[xavier]

You can't get the mates file from anywhere, but have to make it. It consists of the two following lines:

library <name> <min_size> <max_size>\
pair <regexp_forw> <regexp_rev>\

For example, if you have an expected 5 kb library, with a range of size between 4.5 and 5.5 kb (experimentally determined), and your sequencing system appends "_F3" to the reads from one end of the insert, and "_R3" to the reads from the other end, then you could make a file with these two lines:

library solid_5kb 4500 5500\
pair (.*)_F3 (.*)_R3\

(The backslash is just to indicate a carriage return in the example, don't write it in your file.)

I'm not 100% sure about it because I could never try it, but it's what I guess from the documentation.

On the other hand, it seems we are having problems with the ace converter provided by Abyss. If we solve it, then I hope we'll be able to use the amos converter to get the input in the Bambus required format.