Wednesday, June 02, 2021

New Clinical Human Genome Speed Record

I proposed last year that there should be a regular racing event for human genomics.  The only real competitor in is this interesting race seems to be Steven Kingsmore's group at Rady Children's Hospital.  I was sent an embargoed press release from Illumina about a new record by that group, which clocks in at 13.5 hours from patient sample to clinical report. A New England Journal of Medicine paper (hence the embargo, ending just before I post this) reports on the advance but wasn't in the packet I received.
Illumina offered me access to some folks to chat about it, but unfortunately I was travelling yesterday and today had a backlog from taking a 4 day weekend (highlight: seeing my mother, who turns 90 this weekend, face-to-face for the first time since January of last year) so I couldn't take advantage of that offer.  So I'm working off the press release and squeezing this in between obligations.  But a few highlights.

The process takes advantage of Illumina's PCR-free library prep kits, which is an obvious time savings since there is no thermocycling.  It also uses the version 1.5 NovaSeq reagents Illumina rolled out last year.  And probably critically, it uses the DRAGEN software, enabled by FPGAs, to quickly convert the raw FASTQ data into alignments and variant calls.

Their case study in NEJM illustrates the exciting value of such an approach.  A critically ill baby was sequenced in the study and this did enable finding a therapeutic angle which enabled the child to be discharged four days later.  For any parent, finding an answer to your offspring's troubles is a small win -- fixing it is a huge one!

The press release notes that this was run in a research environment and represents a very carefully executed demonstration.  But the same technology is now used at Rady to routinely turn around results in less three days.

Rady's rapid sequencing effort has been used on over 2,000 patients, with 1/3 of those having an diagnosis,  and 2/3 of those -- or 2/9 (22%) overall -- being actionable and one quarter of those (so  5.5%) seeing significant clinical improvement.  That last number definitely indicates we are in early days -- but hopefully the variants seen in the remaining patients are driving new research directions that will increase all of these numbers significantly.

1 comment:

David Eccles said...

If this is from sample to report (and surgery, for that matter), ignoring any measures of genome completeness or coverage, nanopore has that beat (91 mins from biopsy to first report):