I've been accused of being both too sceptical and too glowing about Illumina's technology over the years I've been writing this blog. It very much depends on who you talk to and which post you might have just read. Believe it or not I don't set out to do anything that write down what I think, good or bad. However I'm writing this post in a purely congratulatory tone after re-reading a paper from 2008 by Timothy Ley and Elaine Mardis from WashU: DNA sequencing of a cytogenetically normal acute myeloid leukaemia genome. This was arguably the first cancer genome as we understand it today, and the supplementary methods section makes for interesting reading.
This AML genome was sequenced by creating single-read (no paired-end back then) libraries from several micrograms of DNA, as replicate library-preps. Each of these was sequenced on 98 runs (784 lanes if my maths is right) to generate 6 billion single-end 32bp sequencing reads (or about 7M per lane). At 3 days per lane (a year on a GAII) and approximately $640 per lane in consumables cost $500,000.
Today we can get the same genome using 500ng of DNA in a PCR-free library prep (with no gels), and run this on one HiSeq X Ten lane to generate 375M PE150bp reads. In 3 days for $1000.100x faster, 500x cheaper, and a whole lot easier for you and me! We've come a long way in a short time.