Need suggestions for Illumina genome sequencing strategy

[balaena]

Hi

I need some advice how to cost efficiently sequence 3-4 invertebrate genomes of around 100Mb. I have some experience from a related genome where we had 2x150bp reads, an insert of 500bp and 80x coverage. I was not involved in the sequencing strategy, I simply got the sequences.

The assembly worked quite good: I got an N50 of 115kb and I could further improve the N50 to 350kb with a secondary assembly guided by a closely related published reference genome.

The guided assembly will not be possible with the other genomes and to get comparable results I would like to improve the Illumina sequencing. Naively I was thinking longer reads, longer insert and higher coverage would be better but I have some difficulties to get an overview what the different instruments are capable off, what the companies offer and what they charge. I mean Illumina PE alone because I don't think we can afford a combination with Mate-Pair or PacBio. This is what I know for now:

- We got an offer for $1,400 from a sequencing company for 2x125 (why not 150 on HiSeq 2500?), an insert of 550bp and 110x coverage. But this doesn't look more promising, right? I guess the higher coverage will just outweigh the shorter reads?

- As far as I know the only possibility for longer reads with Illumina are MiSeq platforms, but these are more expensive because of the lower output, right?


Any suggestions for an affordable (around $1,500 per 100Mb genome) Illumina only strategy ?

Thank you!

[HESmith]

Have you considered synthetic long-read (i.e., Moleculo) libraries for improved assembly? The protocol works much better than mate-pair preps, and the reagent cost per library (~$125US) is not much more than for standard Illumina libraries (~$50).

Re: HiSeq 2500 output, 2x125bp is the maximum for older (or upgraded 2000) instruments.

[AllSeq]

Have you considered synthetic long-read (i.e., Moleculo) libraries for improved assembly? The protocol works much better than mate-pair preps, and the reagent cost per library (~$125US) is not much more than for standard Illumina libraries (~$50).

Have you gotten Moleculo to work well? We haven't heard too many success stories.

[HESmith]

It worked better than the mate-pair libraries :-).

In our experience, the critical steps were 1) using input gDNA of the specified size and concentration, and 2) accurate size selection.

Note that our test case was C. elegans, which has a comparable genome size (100Mb) but very little heterozygosity, repetitive DNA (and the repeats are small), or biased G/C content. YMMV.

[AllSeq]

It worked better than the mate-pair libraries :-).

So it was less painful than hitting yourself in the head with a hammer - good to know ;-)

But seriously, thanks for the feedback on your experience with Moleculo - glad to hear it's working out for you.

[kmcarr]

So it was less painful than hitting yourself in the head with a hammer - good to know ;-)

But seriously, thanks for the feedback on your experience with Moleculo - glad to hear it's working out for you.

Has anybody had any experience with the the 10X Genomics GemCode product? It appears to be a small, dedicated microfluidics robot (I'm guessing about the size of a toaster oven) and kitted reagents for creating "Moleculo style" libraries.

When Illumina first presented their kits for creating Synthetic Long Read (neé Moleculo) libraries the protocol looked very long and cumbersome; too many hands on steps. If this little box could do the same thing with a couple of drop in reagent cartridges, push of a button and walk away that would be very interesting.

NOTE: 10X Genomics is being sued for patent infringement over this technology by RainDance and U of Chicago.

[luc]

I would suggest to run all four PCR-free libraries with 2x 250 bp reads on a single rapid Hiseq 2500 lane lane and use Discovar DeNovo.
If you liked the 150 bp PE data, these can be generated cheaper on the Hiseq 3000.

There is no software available (yet) which uses the 10x GemCode data for scaffolding.