Frida, February 12, 2016

James Hadfield-AGBT Guest Blogger (The CRUK Cambridge Institute Genomics Core)

Todays plenary is pretty exciting with two people speaking who have had a massive impact on Genomics, but from outside the academic circle: Jay Flately and Anne Wojcicki. Jay Flatley needs no introduction to the AGBT audience; his driving force in Illumina is one part of what got us to where we are today. His talk titled “Beyond the $1000 Genome – What’s next for NGS?” probably did not give attendees much hope of finding out about much that was new, however Jay delivered in spades. He started by describing some of the key ingredients he thought had driven Illumina, and gave us a brief history of key developments in the Illumina NGS family.

2007 Genome Anlayser; 2010 line scan imaging (HiSeq 2000); 2011 LED optics and Basespace (MiSeq); 2012 rapid run mode, faster fluidics and scanning (HiSeq 2500); patterned flowcells (HiSeq X); 2014 2 colour SBS, dry flowcells and consumer optics (NextSeq).

The exciting bit of the talk was Jay’s very open description of Project Firefly, which he said may not be available till Christmas 2017! This is built out of the Avantome CMOS technology that Illumina stopped developing due to the requirement for emulsion PCRand. Jay showed “the sequencer that was built but never sold”! However Illumina has now implemented SBS on a CMOS to create a one-colour SBS semiconductor sequencer. The system also includes an integrated digital fluidic library prep using the NeoPrep fluidics (not sure whether this is a good thing or not).

Firefly Specs: Raw read accuracy of 99% is achievable and is roughly comparable to HiSeq X, run times are just 3.5 to 13 hours for up to 2x150PE, 1Gb per run, 4M reads at $100 per sample and about $30,000 capital costs. Applications include: targeted sequencing, infectious disease, inherited disease, and reproductive and genetic health.

Anne Wojcicki, asked us to focus on the benefit to the consumer of the Human genome and spent most of her talk speaking about the 23andMe research platform. This now 1.2 million consented customers who can opt-out at any point, or for any research project survey, with all data aggregated and anonymised. 23andME want to get to tens of millions of individuals.

Everything is easily accessed online without geographical barriers and with very rapid participant response. 23andMe have looked carefully at the quality of self-reported data and see high agreement to medical records. The data is high quality and Anne pointed to her favourite paper: Efficient Replication of over 180 Genetic Associations with Self-Reported Medical Data. 75% of 23andMe customers take at least one survey and they have over 320 million phenotypic data points. These are huge communities e.g. 240,000 APOEe4 carriers, 75,000 cancer patients, 34,000 psoriatics, 157,000 with depression, 188,000 cardiovascular disease patients. This is an engaged community with almost 40% of customers still logging in after 6 years, and once they are logged in customers are engaged and this engagement enables discovery.

23andMe therapeutics: can they take a new approach to drug development? The main tools are GWAS + PheWAS and Anne presented an example for target identification in Asthma. 71,000 customers contacted and 6,000 participated to take a six month symptom study. The study costs were negligible compared to normal research trials – “this is going to be transformative in how we can make discoveries”. But she made the case that we need to change how we look at research participants, we need to return results to engage them in the science

Anne did not mention the selection bias of participants and the demographics of 23andMe customers (85% are from the USA). Sharon Plon brought this up in the Q& A, and also the fact that many potential participants are not necessarily online.