metagenome scaffolds

[bsp017]

Dear all,

I have 3 illumina metageome sequence files of the same sample, with different insert size in each. Insert sizes are 300, 400 & 600 bp.

What is the best assembly strategy? Should I merge the files using FLASh, then assemble the merged files using metavelvet or Raymeta? Or should I assemble them individually, then scaffold them using something like Bambus2?

Thanks

[Brian Bushnell]

How long are your reads? You won't be able to merge them unless they overlap.

Ray does a good job of metagenome assemblies using both the merged and unmerged reads together. I've never tried metavelvet. For merging, I suggest you use BBMerge over Flash, as it has a substantially lower false positive rate.

[bsp017]

Thanks Brian,

They are 100 bp paired end reads.

I assembled all 3 libraries successfully using metavelvet. But I have 3 libraries of the same data with different insert sizes, so I want to combine these libraries to get the most contiguous sequence. I tried to input all 3 libraries into metavelvet in one run and the program failed.

[Brian Bushnell]

If you have 100bp reads, and the insert sizes really are 300, 400, and 600bp, there's not much point in trying to merge them, as only pairs with insert size under 200bp could possible be merged.

[bsp017]

OK, Do you think I should scaffold the assemblies?

I had a look at Bambus2 but it wants mate pair data as a pre-requisite.

[Brian Bushnell]

You may be able to do some useful scaffolding with the 600bp library; it's worth trying. I don't know which scaffolding programs are best.