$1000 genome might be actually really close

So I just went to a seminar/marketing plug for this thing, and I must say that it comes across as very impressive indeed. Seems like the cost saving comes largely from just piggybacking onto the currently cheap cost of microchip manufacture to use as the wells for the individual reads. The maximum coverage will indeed be whole human genome rather than just exome with the one they're releasing in six months.
For my money though, I was most impressed by the ability of the system to use its ridiculously large capacity to "barcode" different parallel DNA samples (e.g. from multiple patients) and run a more limited panel of e.g. frequently mutated cancer genes or hereditary disease genes for multiple people on the one chip at the one time with potentially very high multiplicity of coverage. That, I think, would more likely be the real public health benefit - sequence the most important bits for heaps of people at the same time for low cost, rather than the whole genome where most of it isn't really telling you anything. Actually I think that's probably the bigger pure research benefit as well.

On the other hand, I am inherently very very sceptical about high-throughput techniques in general, since they never seem to even come vaguely close to the accuracy the fancy marketing claims. The standout thing that worried me here is that while the actual read is with the fancy new microchip system, the pre-sequencing prep with the emulsion PCR and whatnot seems to be the same as the other "cheap" high-throughput sequencing systems. And that, as far as I'm aware, seems to be the main spot where you get the unevenness and all the potential problems because that's not nearly so precise and neat. So even with the example that they used, even when they budgeted an average coverage of 2000x multiplicity of coverage, almost 5% of the targeted regions had lower than 100x multiplicity of coverage (and it wasn't stated how many of these were complete failures, either). So either you go for massive overkill of multiplicity to get the necessary precision (and to be fair, the precision of the actual read seems very impressive) or it seems you may still miss crucial regions.
Also, the cost was $1000 for the chip itself - what wasn't covered was the cost of all the ancillary reagents and primers and etc etc, which I imagine could be rather considerable. I imagine that someone running a sequencing service based on it would be charging quite a lot more than $1000 to cover all the other associated costs of getting the sample to the machine. To be fair though, the machine itself was $200,000 which is actually pretty damn cheap for that sort of equipment.

So yeah I'm not 100% sure, it's definitely a big and impressive step; whether it's the huge step that's being claimed I'm not so sure about.

Absolutely, rapid screening of large groups would really benefit me. My research is about personalised medicine, and we're targeting specific polymorphisms for intervention. Figuring out if an intervention's effectiveness varies based on polymorphisms would rapidly allow us to select patients for future therapy.

Thanks for reporting.