Blog @ Illumina
Real scientists. Real commentary.

AGBT Standouts

Amy Cullinan, Ph.D.
| Feb 27, 2014

This year marked the 15th meeting of Advances in Genome Biology and Technology, and as every year, kindled the sparks of genomic innovation. Notable plenary sessions included a canny retrospective on sequencing’s pace from Jeff Schloss of NHGRI,  and an update from Dan MacArthur on one of many powerful, practical efforts on exome annotation that continue this momentum. An entertaining and enlightening talk from Tom Gilbert of the University of Copenhagen hailed the entry of a somewhat new topic, environmental sequencing, and everyone in the room, on Twitter, and possibly all of Marco Island  was transfixed  by UCSF’s Joe “Dr. House” De Risi with an amazing story of the ~400 MiSeq reads that saved a life.

Dr. Jeffrey Schloss, director of the National Human Genome Research Institute was certainly speaking to the right audience with his look back on the decade’s progress of next-gen sequencing technology. Beginning with the initial NIH request for applications, Schloss went on to describe the “family tree of NGS”, noting which branches of technology flourished and which went extinct. Several themes were apparent from his talk, including just how early on the concepts for nanopore sequencing had been envisioned, persistent challenges for data analysis still recognizable today, and how formal grant programs, prizes, and technology meetings like AGBT fostered extremely successful collaborations, speeding genome technology evolution faster than anyone could have imagined in 2004. Beyond the now-iconic graph of the cost per human genome sequence outpacing Moore’s Law, Schloss continued to inspire with his thoughts on where advances in genomic technology could take us in 2020.

The genomic tours de force were not in short supply at AGBT; a particularly exemplary one was illustrated by Dan MacArthur’s work from Massachusetts General Hospital on functional annotation of 50,000+ human exomes. MacArthur presented powerful joint calling approaches to generating consistent variant calls across population-sized data sets. The large case and control studies included custom arrays and new compressed file formats, and produced the largest collection of protein coding variants to date. This massive study highlighted the challenges of large scale genomic usability and the benefits of consistently processed data for detecting and annotating rare disease-causing mutations in both complex and Mendelian diseases.

Thomas Gilbert from the University of Copenhagen’s Centre for GeoGenetics gave a brilliantly executed and fascinatingly wide-ranging talk on the use of environmental DNA (eDNA) to assay ecosystem biodiversity. Gilbert’s study focused on threatened mammal species; of an estimated 5,500 known, 40% are threatened by habitat loss, and 25% are under threat of extinction. Tried-and-true conservation efforts are effective, but only when guided by essential genomic data, which can be challenging to find in complex environments such as the rainforest. Gilbert then took off on an anecdote-laden tour of ingenious methods for finding persistent mammalian DNA in substrates such as leech blood, soil soaked with elephant urine, permafrost, bat guano, illegal bush meat markets, and vulture stomach contents. Gilbert touched on the very real barricades faced by chronically underfunded conservation efforts, but left us with a sense of awe that NGS has the potential to aid true species “de-extinction”.

But my very favorite talk of AGBT 2014 was given by Joe DeRisi from the University of California at San Francisco on a critical pediatric encephalitis case.  A 12 year old child from Wisconsin presented with fever and worsening meningitis-like symptoms, and also happened to be severe combined immunodeficient (SCID). A battery of tests failed to reveal an infectious source, and the attending physician had to decide whether to put child on immunosuppressant therapy, broad spectrum antibiotics, or antivirals. In the absence of definitive pathogen evidence, the child’s doctors chose immunosuppressant therapy, resulting in a dramatic crash requiring a medically induced coma and shunts to drain fluid from the brain. As a last-ditch effort to save the child’s life, blood and CSF samples were taken and sent to the DeRisi lab for RNA sequencing using MiSeq. Within 48 hours of receiving the sample, researchers mapped 400 reads out of 8 million to the gram negative pathogen Leptospira borgpetersenii using the crowdsourced project uBiome’s algorithms. A round of penicillin later, and the infection picked up from a lake in Puerto Rico was gone, and the kid achieved a full recovery, just if it were scripted from the TV medical  drama “House”. Confirmatory PCR tests for leptospira that the boy had initially had in his rounds of screening came back negative, but these were known to be only 60% sensitive. Subsequent testing using more amplicons confirmed the NGS results, and begs the question of just how far away we are from routinely using sequencing on infectious disease patients.