While I am here, I though I’d take the chance to chat to the people at the booths for the three major Second Gen sequencing platforms (Illumina, SOLiD and 454). It was surprisingly fun, the guys I talked to all seemed enthusiastic, and it was nice to find out where the scientists in the companies think the technology is going.
In the interests of openness: the 454 booth gave me a cool T-shirt and poster, so this may well have biased my opinion of them
The Illumina guy seemed to mostly want to talk shop. He seemed proud of the 100bp paired end reads, and he thinks they can get 100-150 in the next few months, and hopefully 150-150 by early next year. They are pushing down the cycle time, with the chemical stage being sped up by 30%, and improvements in imaging speed. We talked about insert size; they are getting 6-8kb inserts now, and are targeting 10kb soon. He noted that, of course, the Sanger has some tricks up its sleeve too, and smiled knowingly (to which I smiled knowingly back, pretending I knew what he was talking about).
I asked what he thought Illumina sequencing could offer once the Third Gen sequencing technologies come online. His answer: generating massive amounts of sequence. He speculated that, this time next year, they may be producing 500Gb of sequence per run. Their throughput is still growing exponentially, and they think it’ll be a while before their technology levels off and the 3rd Gen technologies can catch up.
The woman I talked to at 454 was awesome. She was a biologist who did ant transcriptomics (Ant transcriptomics, how cool is that?). She talked about how useful having long read lengths was for transcriptomics, and how she didn’t fiddle with amplification or normalization, instead relying on 454′s ability to use small amounts of DNA efficiently. She said she wanted to see whether they could get 454 to match the low-expression transcriptomics that Illumina can manage.
I asked what 454 will do when new technologies outpace their read length. She took me over to a 454 machine; a white, Mac-ish box, much smaller than the other sequencing machines, with a built in Mac. Sequencing, assembly, filtering, SNP calling, and even CNV and indel calls are all done in the machine, and the pipeline can be extended and played with in the normal unix fashion. 454 are also concentrating on building cDNA and pulldown prep easily into 454 library prep. The idea is that 454 will be easy to use; people will buy 454 not because it generates the most sequence, or the longest reads, but because they can plug it in and go, without any bioinformatics knowledge. This seems like a good plan, and I think it’d be cool if, once again, 454 adapts to new circumstances without going obsolete.
Another thing is that 454 seemed secure. On many occasions she said “well, you should probably talk to Illumina about that”. Also, they gave me a t-shirt and a poster, so I was happy.
No sign of the SOLiD 4 yet, but the SOLiD 3+ is pumping out 50-60Gb per run, with their 2-base ligation sequencing keeping the accuracy up. The SOLiD guy talked about how all analysis could be carried out on the machine, with a built in windows computer and a blade linux cluster handling everything (unlike Illumina, which requires everything to be moved off for analysis). And it is supposed to be more open, with the ability to easily track every read from every bead, without any covert filtering (I wasn’t aware that Illumina made tracking filtered reads that difficult, but ok).
I really want to like SOLiD. They do some seriously cool things; the phasing-by-paired-end technique from their whole-genome paper is very cool, and they call a very impressive array of structural variants. They have the potential to sit in the “Doing Interesting Things In A Cool Way” section of my heart, along with 454. But the guy was constantly saying ‘we are virtually the same as Illumina’, while making sure he pointed out their lack of in-box analysis, their higher false positive rate, their lower read accuracy. Neither Illumina nor 454 made any attempt to talk down the competition, and it just makes ABI seem petty.