454 metagenmic data assembly

[plumb]

Our data is 454 pyrosequencing metagenomic data (not 16s rRNA). We need to assembly them together for metagenomic study. Some people suggest I should try the same way as Transcript assembly instead of genome assembly because the gene does not occur uniformly across genomes. What's your suggestions.

We have newbler, velvet, mira available. Which assembler work the best according to your experience for metagenomic data assembly?

[gridbird]

For 454 pyrosequencing metagenomic data, you should use newbler.

[raw937]

Metagenomic assembly?

Newbler is super slow!!

Mira, isn't great if you barcoded your samples...

I would try Celera or the NEW META-velvet...

[gridbird]

Newbler is super slow if you can not use correct parameters.
for example, "-large" for higher abundance data.

[raw937]

It depends on the read length!
Newbler is a good starting point...

IS it MDA amplified?

I use meta-velvet for the short reads under 100 bp,
CAP3 for the singlets from Newbler
Mira is good for 454...But

depends on what you want?

[plumb]

It depends on the read length!
Newbler is a good starting point...

IS it MDA amplified?

I use meta-velvet for the short reads under 100 bp,
CAP3 for the singlets from Newbler
Mira is good for 454...But

depends on what you want?

Do you mean you run newbler first, then run meta-velvet for the reads shorter than 100bp, and cap 3 for singlets. How do you combine them together?


Also, do you have any suggestion about the metagenomic data cleaning specific to 454?

Thank you

[raw937]

The first thing, I would do is try MG-RAST on your reads.
PrinSeq-lite is a easy perl script to clean your sequences or you can have MG-RAST do it for you off you SFF files.

Second,
Newbler first for your reads.
The singlets should be assembled with CAP3.

Meta-velvet is best for illumina reads..

Let me know what you need scripts etc?
I have been in analysis hell trying to figure all this out, I GOT it so there is no point for you to have the same problems...

LET ME KNOW
RAW937

[plumb]

Hi Raw937

After quality control, I have assembled the data by using newbler 2.5. 50% reads assembled into contigs. Then I tried cap3 by using default parameters, around 50% singletons assembled into contigs. However, those contigs are short. only 5% of them are longer than 900bp. What's your suggestion for the CAP3 parameter? Why you think we should use CAP3 here not other assembler? how about newbler with less stringent parameter? Have you thought of mapping assembly?

Thank you very much!

[raw937]

assembly

I do both newbler first to get the main contigs.
Then CAP3 for the singlets from 454.

You can map you CAP3 only contigs to you Newbler contigs to validate who is telling the truth.

I would get an MG-RAST account it will help you out alot to tell you who, what and how your community functions.

Its very fast and you can compare to other metas very easily.

I would use prinseq lite perl script to clean, then post your reads...

Your in Canada eh?

I am at UBC where are you?

Thanks
Raw937