Blog @ Illumina
Real scientists. Real commentary.

AGBT Day One and Illumina User Meeting

Abizar Lakdawalla, Ph.D.
| Feb 14, 2014

The 15th AGBT meeting started on Tuesday evening, but many attendees had problems reaching Marco Island due to weather disruptions. Thus, the main hall during the first plenary session looked less filled than it could have, in spite of the >850 registered attendees.

Illumina held its user meeting on Tuesday Feb 12th before the official start of the conference, attended by ~200 people. As previously mentioned, Illumina's R&D folks detailed the substantive changes in the engineering designs, NextSeq chemistry, and patterned flow cells for the HiSeq X Ten (accompanied by lots of engineering graphics from SOLIDWORKS and SEM images of flow cells).

Frank Oaks described the early phases of NextSeq development, including engineers taking apart DVD and Blu-Ray players to finally end up with the new NextSeq cameras, which are fully sealed. Six of these small cameras are used for speed, and in combination with a new monolithic fluidics architecture, make the NextSeq a highly reliable box. The NextSeq underwent an inadvertent robustness test, when it was dropped and still worked after being sheepishly picked up from the floor and placed on the bench. Real data from whole human genome sequencing was shown for the new chemistries and compared with data from the HiSeq 2500 platform with relatively identical results. The HiSeq X (“comes in handy 10 packs”) also has a new camera with dual direction scanning, wider field of view, higher power lasers to give a 6x increase in scan speeds. Also has a beefed-up internal computer to handle the higher data rate.

Vince Smith showed the modified sequencers (GA, MiSeq and NextSeq prototypes) that his group used to optimize the new 2 channel SBS chemistry which speeds up the system by using just two images per cycle. Base calls are read as binary codes – 1:1 = A base, that is, red:green = A; 1:0 = C (red only); 0:1 = T (green only) and: 0:0 = G (no fluor). The miniaturized cameras in the NextSeq required bigger and brighter clusters that led to the development of a new surface coating chemistry for the NextSeq flow cells, which also allows them to be shipped dry.

Johanna Whitacre from R&D talked about the development of the patterned flow cells for the HiSeq X which increases the data rate by 10x of the HiSeq 2000 for an effective increase in 30x human genomes to ~1,800 per year per HiSeq X as compared to about 170 per year on HiSeq 2000. The patterned flow cell with 400 nm wells spaced on 700 nm center-to-center spacing currently produces about 1,600,000 pf DNA clusters per mm2. Cluster generation is based on a new exclusion cluster chemistry, an interesting side effect of which is that cluster density is relatively immune to higher library input amounts.

James Hadfield from Cancer Research UK pulled no punches by starting saying that he had found BaseSpace “underwhelming” until test-driving the new RNA-Seq App which he found genuinely useful and easy to use. He was surprised at the ability to easily produce publication-quality graphics from the BaseSpace App, albeit “probably for PLoS One rather than Nature”.

After the Illumina user meeting, there was a bit of a rush to the Illumina lounge where a couple of NextSeqs and a NeoPrep, the new automated library prep device, were subjected to inspection. The large flow cell of the NextSeq was scrutinized closely by most folks hoping to see the bigger clusters with their naked eyes. The simplicity of the NextSeq workflow combined with a 120 Gb output continues to surprise. The raw 8-inch wafer containing the patterned flow cell surface and the fully assembled HiSeq X  flow cells sporting eye-catching diffraction patterns from the 400 nm wells were admired. James Hadfield even managed to get Jay Flatley to autograph one of the HiSeq X flow cells so that he could take it back with him.