This was not a blatant omission on our end.
At the time of submission the ILMN genotypes were not available. They posted the coordinates of their variation but not the actual variant genotype making a direct comparison weak and probably uncredible. Bit odd that a Nature paper can exist without such critical information being available and I dont see any reason why they would choose to keep these private for such a long period of time.

We have no such reservations and plan to have all variants public.

- it is the policy of Genome Research to withhold the supplemental data until the paper is in print

My subjective 2 cents on Niche and dodos….I dont think this race is over in the 1st 900 machines considering over 12,000 CE instruments have been sold. Most analysts are predicting this market to be far bigger than the CE market and predicting SOLiD will have a part of it. For what little its worth, ILMN is down ~20% this week based on missing their numbers. LIFE is up.

Additionaly, Just as Bentley et al was GAI data this is all SOLiD V2.0 data in the paper. The field is moving quickly and 50Gb runs are happening quite frequently on SOLiD 3.0s (One presented @CSH this 5/09). Many would argue SOLiD has been the pace setter and has a history of always leading the throughput curve in terms of Gb/day.

ie..CSH Biology of Genomes meeting has for the last 2 years had SOLiD presentations from customer with 2-3X more data/day than ILMN with reports of 99.91% accuracy (5-10X higher than the reported accuracy on ILMN at the time).

Obviously, mate pairs are not unique to SOLiD but the accuracy is and it enables the investigation of phasing (must be confident you have 2 SNPs in your mate pair, not a SNP and an error). If you have paired 50mers (100bp) and the reported >1% error on ILMN, you will always have at least 1 error in your paired end. Makes a phasing study a bit more cumbersome. We have not seen others phase structural variants like we have with SOLiD. These long insert high accuracy reads will become of increasing value as we move to more reference free de novo assemblies.

I dont think there is a Next Gen genome done with this much physical coverage and insert size. Nor do I believe other platforms have yet bridged the gap between small indels and large indels to this extent. Of course, there are several SOLiD users with far deeper genomes than this now so its an emphemeral title:)

Are both datasets in the Short Reads Archive? It would be a great project for a grad student or advanced undergraduate to compare the two sequences. Some of the contract vendors I have spoken to claim the SOLiD has lower error rates, and it would be great to get some hard numbers on that.