How two compare two assembly results? (not assembly tools)

[strob]

Hi,

most of the time, people compare results, using the same dataset, but obtained from different tools (example: velvet vs. soapdenovo)

I have however two independent datasets (solexa data, obtained from two different 'extraction' methods), on which I performed seperately a de novo assembly. So I have two assemblies, coming from different data, but coming from the same species. Now I would like to compare both assemblies and see which assembly is the most satisfying to me. I have questions like: is there overlap? Is one completely imbedded in the other? Are they complementary? etc.

So simple: how to compare two independent de novo assemblies (same species)?

Thanks

[nickloman]

I would use Mauve to do this and visualise the output. If you had any reference-grade genomes that you would expect to be similar and preserve gene order (synteny) you could also add that in to guide which assembly is likely to be more accurate.

[maubp]

There is no easy answer - and it depends on what you are looking for.

For example, further to Nick's excellent comments, you might want to run a gene caller on both and compare the gene sets. If you are interested in particular genes, then you could just try TBLASTN or BLASTN to see if they are in both draft genomes or not.

[strob]

I used two different methods to extract my DNA, which was than sequenced and assembled. I want now to know which method gave me the best results, by comparing the two assemblies. I want to find out if both methods give me the same data or if there is only a partial overlap.
Mauve is nice to visualize, but limiting. I would more like to get some overall statistics

Thanks already for the feedback

[nickloman]

After you've generated a Mauve alignment you can use Mauve Assembly Metrics (http://code.google.com/p/ngopt/) to give you some stats, but this is designed to be used against some kind of baseline truth reference genome. But you could run it twice I guess with genome 1 as reference first time and genome 2 as reference second time in order to generate some numbers.