Jason Betley from Illumina R&D, started off with an introduction and reviewed some recent products. The HiSeq 2500 can sequence a human genome in less than 24 hr due to a simpler sample prep (PCR- and gel-free) and the new ISAAC aligner can handle a human genome in about 2 hours. The PCR-free workflow demonstrated on cell-free DNA from maternal plasma at 200x coverage was shown as an example. Betley mentioned that read lengths on the HiSeq 2500 will increase to 2 x 250 bp in the second half of 2013, giving 300 Gb in 2.5 days.
On the MiSeq, 2 x 300 bp on 25 million clusters would also be available later this year, as previously announced. Betley provided more details about the TruSeq Targeted RNA sequencing assay for quantitative gene expression, giving data in 2 days at a lower price than qPCR. The Nextera Rapid Capture exome kit was also mentioned, allowing exome-level experiments to be run on MiSeq from 50 ng of gDNA with 96 indexes.
Stephan Schuster from Penn State, presented a lively talk on the value of longer reads for de novo assembly and metagenomic analysis. As an example, a library from sewage sludge was used to generate 92 Gb of 400+ bp overlapping reads from 2 lanes of the HiSeq 2500 with a 0.6% error. The longer read length gave a greater certainty in aligning and identifying organisms with MEGAN software. With just two lanes of sequencing, saturation for organism identification was apparent.
Mauricio Carneiro from the Broad Institute wrapped up the night with a discussion of the challenges of human genome sequencing. The Nextera Exome protocol showed tighter coverage distribution, unlike their current Agilent protocol that showed a big peak close to 0-1x coverage and a long tail with excessive coverage. The Nextera protocol covered targets that were not previously covered, and seemed to perform better in GC regions. For detection of variants at 99% sensitivity, only 8x coverage is needed for het calls, but for simple indels, 39x coverage is required. False negatives decreased with increasing read length. For accurate indel detection, read length is absolutely critical, true indel positives increased from 23 to about 230 at longer reads. They've also tried 2 x 400 with 85x coverage of the human genome on the HiSeq system, which generated some excited tweets from various audience members!