[AllSeq]

AFA I can tell NextSeq 500 does not use ordered flowcells and the two color chemistry is only for that instrument.

Yeah, the ordered flow cells appear to only be associated with the new HiSeq X machine (at least for now).

[GenoMax]

Do they really need the five first bases to identify clusters on an ordered flowcell?

AFA I can tell NextSeq 500 does not use ordered flowcells and the two color chemistry is only for that instrument.

[Chipper]

Do they really need the five first bases to identify clusters on an ordered flowcell?

[TonyBrooks]

So if you have 5 Gs at the beginning of a template, there will be no cluster found?

Would not a bubble in the flowcell result in all those clusters being called "G"? (At least for the bottom surface.)

Surely anything beginning with 5 G's won't be registered as a cluster. So bubbles wouldn't contain any clusters either - so you don't get any data. I would expect that there is some algorithm there to detect when something is a G and when it's an intermittent bubble. G's would be localised areas with no signal, bubbles would cover much larger areas.

You'll end up losing less than 0.1% of your clusters assuming total randomness (1 / 4^5) but I guess that's a hit worth taking to effectively half the imaging time.

One thing I'd like to know is whether NextSeq employs the empirical phasing calculations used on the MiSeq, or whether that's too computationally expensive.

[pmiguel]

What I'd like is some more detailed explanation of the 2-dye system the NextSeq 500 is using. I assume they provide C & A tagged with Red, T & A tagged with green & G untagged, thus C should be pure Red, T pure green and A ~ a 50:50 mix of green & red (orange) and G then undyed...

So if you have 5 Gs at the beginning of a template, there will be no cluster found?

Would not a bubble in the flowcell result in all those clusters being called "G"? (At least for the bottom surface.)

[TonyBrooks]

The rep at PAG said they planned on selling 4 "X boxes" in 2014--and had already sold 3 30 minutes after the announcement.

The NextSeq500 is clearly aimed at people considering a Proton. Same price, same emphasis on speed, better stats right now. I don't like that the only single-end mode is 75 bp. Lots of people like the 150 bp read on the 2500 Rapid. And the other downside is that all 4 lanes (100M reads each) get fed the same library, so doesn't improve flexibility in flow of projects at a facility. But a nice mid-machine.

There is a 150 cycle kit. It's marketed as 75 paired end, but if 150 read length is supported, I'd we surprised if you're not able to do a single 150bp read (just like you can use the MiSeq 150 v3 to do 150 single read or 75bp paired end).

[ymc]

The restriction is that the minimum purchase is 10 machines at a time (but they don't need to be purchased only in 10s - someone had an initial order of 14). There has been nothing to indicate that the instruments operate together - it's just a minimum purchase defined by Illumina (essentially for pure marketing purposes - if you can't afford 10, you aren't in the right customer segment for this machine). Same logic behind restricting it to running human samples - pure marketing choice.

That makes sense now. I suppose that price might also include the salary of a local support person. I know they hired several support guys in Hong Kong when BGI bought the 128 machines.

[ymc]

Really, this is a bit of a giant FU to Complete Genomics and, uh, BGI.

Yeah, this HiSeqX has 10x throughput over HiSeq 2500 SBS v3. Obviously that's a very big FU for BGI indeed.

But I heard that they have more management issues than technical issues.

Anyway, if Dr Yang can convince the right commies again to buy a bunch of HiSeqX Ten, he can still be in the game. The NYT news today said China printed more money than the US.

[AllSeq]

There is still one thing I don't quite understand about HiSeqX. That is: do you need to buy it in 10-pack? It appears to me Table 1 in their data sheet shows the stat for one box. It doesn't seem to me you need 10 boxes to run together to archieve that throughput.

Is it because if you buy it in 10-pack, then you get a significant hardware discount such that the lower instrument depreciation cost can push the cost of 30x genome to be under $1,000?

The restriction is that the minimum purchase is 10 machines at a time (but they don't need to be purchased only in 10s - someone had an initial order of 14). There has been nothing to indicate that the instruments operate together - it's just a minimum purchase defined by Illumina (essentially for pure marketing purposes - if you can't afford 10, you aren't in the right customer segment for this machine). Same logic behind restricting it to running human samples - pure marketing choice.

[ymc]

There is still one thing I don't quite understand about HiSeqX. That is: do you need to buy it in 10-pack? It appears to me Table 1 in their data sheet shows the stat for one box. It doesn't seem to me you need 10 boxes to run together to archieve that throughput.

Is it because if you buy it in 10-pack, then you get a significant hardware discount such that the lower instrument depreciation cost can push the cost of 30x genome to be under $1,000?

[aeonsim]

I agree this is a stupid restrictions. However, is there a lot of drive to do huge scale, medium depth, WGS on a species other than human? How many 30x mouse genomes can one need?

You aren’t going to get a de novo genome with 30x at a single fragment size, at least not one worth much. And the de novo sequencing must make up a tiny fraction of sequencing projects in the world. So, if Illumina is going to have some 5 different machines now, what the problem in having one machine dedicated to large scale human resequencing studies? Maybe the implementation is wrong, but I think the general idea is right, especially from a marketing perspective.

At least with Cows there are multiple groups sequencing as many animals as they can afford to get a better understanding of variation within and between breeds. Even with the previous prices the groups were sequencing between 140-700 animals each (plus a community 1000 Bull genomes project), if the price went down we'd sequence more and apply for more funding (both industry & public) to get a decent understanding of the different breeds and the range of variation in the population which is critical for animal breeding etc

[westerman]

I agree this is a stupid restrictions. However, is there a lot of drive to do huge scale, medium depth, WGS on a species other than human? How many 30x mouse genomes can one need?

I suspect that doing GWAS studies on mice, dogs, etc. could be of interest. It is nice to to have inbred lines but being able to sequence the outbred lines could produce interesting results that can be easily experimentally verified in ways that human experimentation is not allowed.

Now funding such a study might be difficult. It is one thing to say "$1000 to sequence yourself" versus "$1000 to sequence your pet".

You aren’t going to get a de novo genome with 30x at a single fragment size, at least not one worth much.

Agreed.

And the de novo sequencing must make up a tiny fraction of sequencing projects in the world. So, if Illumina is going to have some 5 different machines now, what the problem in having one machine dedicated to large scale human resequencing studies? Maybe the implementation is wrong, but I think the general idea is right, especially from a marketing perspective.

Yes, I think that is a good point. A year from now after the initial sales have slowed down then perhaps marketing will point towards other species.

[Wallysb01]

Read that, it provides the basic information but not much more.

The one really stupid and highly annoying thing about the Hiseq X is the restriction to only sequence Human genomes (at the $1K 30x cost). It's an entirely arbitrary restriction to increase there profits and nothing else.

Especially seeing it sounds like it's a software based restriction which effectively means they're selling you hardware (expensive hardware) that they've purposefully crippled if you want to do anything other than Human genomes. Highly aggravating hopefully some of the groups that buy the instruments will come up with a work around or means of disabling this restriction, certainly the first service provider who can get a large (~3GB) non-human genome out for ~1K will have plenty of interested buyers (assuming Illumina contractually doesn't prevent this...).

I agree this is a stupid restrictions. However, is there a lot of drive to do huge scale, medium depth, WGS on a species other than human? How many 30x mouse genomes can one need?

You aren’t going to get a de novo genome with 30x at a single fragment size, at least not one worth much. And the de novo sequencing must make up a tiny fraction of sequencing projects in the world. So, if Illumina is going to have some 5 different machines now, what the problem in having one machine dedicated to large scale human resequencing studies? Maybe the implementation is wrong, but I think the general idea is right, especially from a marketing perspective.

[aeonsim]

Spec Sheet from ILMN:
http://bit.ly/1iOYN7e

Read that, it provides the basic information but not much more.

The one really stupid and highly annoying thing about the Hiseq X is the restriction to only sequence Human genomes (at the $1K 30x cost). It's an entirely arbitrary restriction to increase there profits and nothing else.

Especially seeing it sounds like it's a software based restriction which effectively means they're selling you hardware (expensive hardware) that they've purposefully crippled if you want to do anything other than Human genomes. Highly aggravating hopefully some of the groups that buy the instruments will come up with a work around or means of disabling this restriction, certainly the first service provider who can get a large (~3GB) non-human genome out for ~1K will have plenty of interested buyers (assuming Illumina contractually doesn't prevent this...).

[GW_OK]

Maybe more "headspace" by driving scanning/optics?

I'm thinking their cluster calling algorithm is not quite as robust and/or they're playing it safe with only two images to work with per cycle.