Genomic sequencing could help cure cancer, but first we must manage the data

The field of genomics is on the precipice of bringing real, tangible benefits to people worldwide, but we need to figure out how to handle all the data it produces, said Francis deSouza, president of Illumina, at Gigaom’s Structure conference in San Francisco on Thursday.

DeSouza pointed out cancer as a problem area where genomic research could lead to big jumps in treatment efficacy. “We classify cancer by where it shows up, instead of the genetic profile of the cancer,”  deSouza said. “That is a primitive approach to the problem.”

DeSouza would recommend a full sequence of both the tumor and the patient as a first step to anyone unlucky enough to be diagnosed with cancer in 2014. A genomic researcher at Washington University in St. Louis, Lukas Wartman, did just that a few years ago when he discovered an off-label use for an existing drug which helped push his acute lymphoblastic leukemia into remission.

These advances are becoming possible because of the rapid decrease in cost associated with sequencing. When the first human genome was sequenced in 2003, the total cost was close to $3 billion. Now a full sequence is as inexpensive as $1000. There’s a lot of technological innovation that has helped drive that price down — advances in molecular biology, optics and semiconductors among others — but as prices decrease, it leads to a lot of data that needs to be secured and aggregated. “One gram of DNA contains two petabytes of info. Just massive, massive amounts of data,” deSouza said.

Not only is there a lot of data, but that data needs to be secured. Currently, the hospital or foundation that does the sequencing owns it. This brings up important questions: Who is encrypting it? Who is protecting it?  “I should own my own genomic data, I feel,” deSouza said. “You want it anonymized, but available.”

But allowing patients to lock down their own data could end up being a tragedy: genomic data is most useful in aggregate, as to pick out trends in certain populations and demographics. Currently, when a study slightly changes its scope, it often has to ask for consent from all the providers of data once again, which ends up being a headache. “We need a consent infrastructure allowing the use of this data,” deSouza said. “Currently consent is very specifically granted, and new studies might need to regain consent.”

If genomics figures out its massive data problems, it could lead to a major revolution in the way we treat cancer.

Photo by Jakub Mosur

Structure 2014 ticker