Tweet
Have you seen the investor press release:
Anyone tried to parse this to get the actual cluster density change?
This would be 300 Gb from 4 lanes, right? (2 flow cells, 2 lanes per rapid flowcell).
So 75 Gb per lane. That does sound like a lot. But much of that would come from the 100 --> 250 base read length shift.
So 30 Gb per lane with 100 base reads. That does not sound so great. Using "high output" chemistry one gets 40-50 Gb per lane with 2x100 PE at optimum clustering densities. Although, I guess that is beyond spec for a HiSeq. But it is what you get.
But, currently anyway, only 2 swaths per lane are scanned with Rapid chemistry, instead of the 3 swaths per lane scanned with High Output chemistry.
So what does this mean for the actual increase in cluster density? 30 Gb at 2x100 base reads converts to 150 million clusters/lane. What is that? ~160K PF Clusters/mm^2 for 3 swaths, so 240Kclusters/mm^2 PF for a 2 swath lane?
That is not bad, especially if quality is good out to 250 bases. However, I don't think it represents a dramatic price per base decrease. Of course that may not be the appropriate metric. 2x250 for de novo applications give you more than 2.5x the sequencing power of 2x100.
I am bringing this up, because we have seen the hyper-Moore's rate increase in bases/$ for several years now. One always wonders when this will top out. My impression, which is purely the speculation of someone with zero inside connections in the industry, someone literally working in a hole in the ground (well, the sub-basement), is that Illumina is deliberately throttling back on the bases/$ acceleration to just what is needed to compete with the Proton Torrent.
For instance, the 2 swaths/lane scanning of Rapid flow cells seems completely artificial to me. The lanes are clearly large enough to accept an additional swath. But this would have the effect of reducing the cost/base to below that of the "high output" chemistry.
Personally, I am completely ready to kick the "high output" chemistry to the curb. I mean, it delivers, but 10 day runs are exhausting. And even minor problems in a run generally result in a week of troubleshooting, minimum, to get a fix. That kind of device period just isn't good for humans. At least if you have a single instrument. If you have several, you can just stagger the runs to get data coming off in even intervals.
--
Phillip
HiSeq 2500 chemistry enhancements empower the industry’s highest daily throughput and drive down the price of whole-genome sequencing. With support of paired 250 base pair read lengths in rapid run mode, the HiSeq 2500 will be capable of generating up to 300 Gb in rapid mode with sample to data in less than three days. These enhancements will be available in the second half of
this year.
this year.
This would be 300 Gb from 4 lanes, right? (2 flow cells, 2 lanes per rapid flowcell).
So 75 Gb per lane. That does sound like a lot. But much of that would come from the 100 --> 250 base read length shift.
So 30 Gb per lane with 100 base reads. That does not sound so great. Using "high output" chemistry one gets 40-50 Gb per lane with 2x100 PE at optimum clustering densities. Although, I guess that is beyond spec for a HiSeq. But it is what you get.
But, currently anyway, only 2 swaths per lane are scanned with Rapid chemistry, instead of the 3 swaths per lane scanned with High Output chemistry.
So what does this mean for the actual increase in cluster density? 30 Gb at 2x100 base reads converts to 150 million clusters/lane. What is that? ~160K PF Clusters/mm^2 for 3 swaths, so 240Kclusters/mm^2 PF for a 2 swath lane?
That is not bad, especially if quality is good out to 250 bases. However, I don't think it represents a dramatic price per base decrease. Of course that may not be the appropriate metric. 2x250 for de novo applications give you more than 2.5x the sequencing power of 2x100.
I am bringing this up, because we have seen the hyper-Moore's rate increase in bases/$ for several years now. One always wonders when this will top out. My impression, which is purely the speculation of someone with zero inside connections in the industry, someone literally working in a hole in the ground (well, the sub-basement), is that Illumina is deliberately throttling back on the bases/$ acceleration to just what is needed to compete with the Proton Torrent.
For instance, the 2 swaths/lane scanning of Rapid flow cells seems completely artificial to me. The lanes are clearly large enough to accept an additional swath. But this would have the effect of reducing the cost/base to below that of the "high output" chemistry.
Personally, I am completely ready to kick the "high output" chemistry to the curb. I mean, it delivers, but 10 day runs are exhausting. And even minor problems in a run generally result in a week of troubleshooting, minimum, to get a fix. That kind of device period just isn't good for humans. At least if you have a single instrument. If you have several, you can just stagger the runs to get data coming off in even intervals.
--
Phillip
Comment