How to improve diploid assembly?

[Poni]

Hi,
I'm a beginner to the field of sequence analysis.
I'm currently working on assembly of diploid fugal genome which is around the sizes of 60Mb. Sequenced using Illumina HiSeq2000. I have used Velvet assembly software to construct contigs. However, it ended up in more than 200,000 contigs with low value of N50 (less than 1000) for Kmer =45. My other assembly for a haploid genome was having 20,000 contigs and N50 around 50,000. Therefore, I'm just wondering how I should improve the velvet assembly in order to improve my assembly. Please may I know if there is any important stuff regarding improving assembly for diploid genome (source where I can get information) , or about any other suitable assembly software for this purpose.

Thank you
Poni

[lexa]

What kind of libraries do you have sequenced? Only paired end libs or also mate-pairs? You can try to use other assemblers (SOAPdenovo, Allpaths-lg, ...) to improve your assembly.

[ians]

Poni,

Likely the issue isn't that it is diploid, but rather you just have a much more complex organism (which is often the case of diploid organisms.) The assembler doesn't really care as it will treat most alleles as "sequencing error" and just squash the allele into a consensus sequence.

There are two huge factors to getting a great assembly:
1) Library diversity: e.g use paired-end libraries with fragment sizes of 300,500,2k and 5k. It's important to get atleast one LIMP library and make sure its devoid of any adapters.
2) Max out your Kmer. If you have the memory for it, juice the Kmer size as high as your Kmer Coverage will allow.

To get across repeats, you need to have large fragment sizes and informatically use large reads with a kmer size that uses as much as the read length as possible. From there, you can tweak parameters.

[Poni]

Many thanks for your response.
I have been sequencing paired end libraries. I already used SOAPdenovo but it didn’t add much.
Thanks for your advice. I will try how to go with a higher Kmer size.

Thanks Poni

[nextomics]

To Poni:

So after you use a higher Kmer size , do you get a better result?