Plant whole genome hybrid assembly

[Hitesh]

Hi,

We have generated data using Illumina HiSeq 2000 and 454 for a plant species.

We completed the hybrid assembly using Velvet and got assembled contigs.

Now we were trying to map the Illumina reads on the assembled contigs but only 25-30% of the reads are getting mapped.

We were expecting more % mapping !!!!

Does any one has any experience with this ?

[jimmybee]

Are you using all assembled contigs to map against? Or did you only choose contigs greater than a certain size?

[Hitesh]

We used contigs greater than 300 bases...

[krobison]

What length Illumina data do you have? What fold coverage given the expected genome size? What is the total contig length & N50 for your assembly?

I would try assembling the Illumina data alone & see what you get. Also, try Ray as your assembler -- it tends to build longer contigs, and my experience to date is that the longer contigs are valid.

Have you scanned different kmer values for velvet, either manually or using VelvetOptimiser?

[Hitesh]

We have generated 45GB illumina data (100PE), 1 full plate 454 FLX and 1 plate 454 FLXplus for approx 700MB genome...we finally got ~400,000 contigs covering 630MB genome with N50 of 7.8KB...we did try different kmer lengths by manual calculations...the problem is that now when we try to map the illumina reads to the contigs ( ~ 400,000) to get heterozygous calls only 25-30% of the reads get mapped which we feel is a little strange !!! Is this expected ?

[jimmybee]

A lot of your reads will map to your singletons (<300bp contigs) so you should try and include those in your analysis to see if the mapping improves. Is the genome repetitive? Have you had a look at the read coverage of the 25-30% that mapped? Is it uniform?