There’s a lot of excitement and a fair amount of drama at AGBT this year. PacBio, the gold sponsor, seems to be flailing. Illumina remains serene in spite of a host of up-and-coming sequencing companies promising the moon and the stars. And we’re all trying to ignore the elephant in the room: a giant IonTorrent tour bus parked out on the beach and the accompanying army of unregistered Life Tech pushers.
The Challenge of Clinical Annotation
The day started off somber enough, with a morning session on “Clinical Translation of Genomics” (chaired by Chad Nusbaum) in which everyone began to appreciate the truly difficult problem of accurate clinical annotation of genetic variation. Sequencing and variant detection are accessible and straightforward to most labs. But interpretation of the results, even for rare genetic diseases, has proven challenging. Lynn Jorde gave a personal touch to his lab’s work on the Miller syndrome family. Heidi Rehm spoke about the need to expand current targeted clinical tests.
The afternoon session, chaired by Len Pennacchio, was an interesting mixed bag. Goncalo Abecasis (Univ. of Michigan) gave an outstanding talk on 1,000 Genomes and the sequencing of complex traits. He and his collaborators found, perhaps unsurprisingly, that genotype accuracy goes up with the number of samples sequenced, and also that the consensus call of multiple variant detection algorithms outperforms any single caller alone. Ewan Birney (EBI) tried to squeeze the massive ENCODE project (30 lead PIs and a main paper with ~350 authors) into a talk that I’m not allowed to blog about (manuscripts under review). Chuck Perou (UNC) gave a different sort of talk on the challenges of vetting new gene expression assays (Nanostring, RNA-seq) in profiling cancer samples whose standard archiving process involves formalin-fixation paraffin embedding (FFPE), a process that systematically shears nucleic acids.
Next-gen Sequencing: Lots of Data, Not Many Findings
The evening concurrent sessions have been well-covered by Twitter feeds (search: #AGBT). For the most part, I stuck around the “Medical Sequencing and Genetic Variation” session, chaired by Elaine Mardis (WashU). The recurrent theme, ironically, seemed to be “We’re not sure what to make of this.” It didn’t matter if the phenotype was super-longevity (Devine), autism (Edwards), or an undiagnosed neurological condition in monozygotic twins (Margulies). The technologies were advanced, the analyses were [usually] thorough, but there just aren’t any definitive novel findings. More work to be done. More samples needed. No one’s blaming the sequencing technology.
And so, while the would-be-overlords of sequencing duke it out over throughput, accuracy, and cost, most of the delegates are really thinking about what’s coming: the daunting task of systematic analysis and interpretation.
[…] has been low on announcements so far. As Dan Koboldt says people are now a bit more focused on how to interpret all that lovely data they’ve been […]