Validating of Pacbio Assembly

[macb]

Hi !
I'm working on an hybrid de novo assembly of a 2 Giga bases genome using Pacbio and illumina Paired End reads.
I would like then to validate my assembly and do some controls such as detecting collapsed repeats, misassemblies and improve the consensus accuracy.
So could you please say me which is the best tool that can do that and is suitable for large genomes like mine.

[rhall]

It is going to be difficult to validate an assembly if the only data you have is the data used for the assembly itself.
What do your current assembly stats look like, what is your coverage of PacBio and Illumina data, N50 of the PacBio reads, insert sizes for your Illumina libraries?
If you can assemble leaving out a specific set of data you could then use that for validation.
For fully denovo validation, you could look at annotations / gene predictions, if you have a good gene model, or something like ALE, but I havn't seen it used on such a large genome.

[macb]

Dear rhall,

My genome assembly is not finished yet, however I can give you some statistics:
-Estimated genome size : 2Gb
- Illumina Paired End : 40x insert size : 400bp
-Illumina Mate pair 5kb : 4x
- Illumina Mate pair 20kb : 1x
- Pacbio reads : 14x with N50 : 8331bp
I have found Pilon from broad institute as a interesting tool, however il looks as time and memory consuming

[rhall]

The mate pair libraries will not add much, if anything to the hybrid assembly, leave them out and use them to test consistency.
Pushing the PacBio coverage to 20X (http://wgs-assembler.sourceforge.net..._Release_Notes) would allow you to do an assembly with only the PacBio data, then use the Illumina for validation.

[macb]

In fact I would use mate pair libraries for for ultra-scaffolding the resulting assembly using SSPACE. Then, I would close gaps using a stand-alone program like GapFiller. I have noted that Pilon closes gaps also in addition to the assembly validation.
Does GapFiller corrects misassemblies ?

[rhall]

I'm not familiar with GapFiller.
Conceptually given the N50 of the PacBio data I would assume that the coverage @ 20kb and 5kb results in higher information content in the PacBio reads than the 4x of 5kb and 1x of 20kb mate pair data. Therefore I would be surprised if the mate pair data added much information in the way of scaffolding, it may be more useful in validation.

[macb]

OK wright. Which tool would you suggest to me for validating assembly with illumina reads.

[rhall]

I'm not too familiar with it myself, but AMOS contains tools that can be used for assembly validation using mate pair data, in particular the asmQC tool.

[macb]

Can I use Paired end and mate pair reads with asmQC tool ?