If DNA code can hold the information for making an entire person, surely it can do the same for an integrated medical record? Such is the reasoning of the developers of the Genomic Messaging System.
Acknowledging that the project began "when the DNA-augmented patient record seemed to many to be mere science fiction," scientists Barry Robson and Richard Mushlin of IBMs T.J. Watson Research Center now predict GMS will be able to transmit information as a "smart DNA sequence" that contains a patients entire medical record along with genetic information. Even MRIs and X-rays might be rendered into the system.
GMS is essentially a computer language that allows clinical information to be embedded within streams of DNA sequence. Management functions, such as security measures, can also be embedded in the data stream. Robson described the transmitted data (with a nod to Hollywood) as "a stream of DNA symbols—GATTACAGATTACA—with GMS language inserted at appropriate points."
"GMS links archives of digital patient records to enable analysis of those records by a variety of bioinformatic and computational biology tools," said Robson. For example, patient data could be more easily mined for interesting correlations, say between an insect bite and pancreatitis.
The researchers originally developed the system for building clinical genomic reports of patient data, then generalized it so that clinical data could be more readily linked to bioinformatic and computational tools.
Most recently, in a proof-of-concept demonstration GMS was used to combine information from a patients medical record in Israel with DNA data from New York. The scientists used GMS to predict how single nucelotide polymorphisms, or SNPs, would affect the structure of a histocompatibility protein for that particular patient, and how that structure would affect the way drugs interact with the protein.
For example, modeling could detect whether an SNP deleted particular sites on the protein, such as glycosylation or cleavage sites, important for the proteins or drugs activity. Such information could one day be used to match an individual patient with the most effective drug or dosage—or even to design a drug specific to the patient, the IBM scientists speculate.
The researchers acknowledge that such work is very far off and that such a problem would need to be solved by pharmaceutical science, not just information science. Still, they speculate that GMS is sufficiently powerful to support personalized medicine, at least for the purposes of information technology.
The code used for this proof of concept was Perl 5 with capabilities for XML management. However, the researchers emphasize that several data types can be "embedded" in a DNA sequence. "The GMS system is larger than, and not confined to, an XML based approach." Moreover, the scientists wrote, GMS is designed to be "minimally dependent on other systems," which could make it ideal in emergency situations requiring portable equipment. "The messaging network could comprise direct communication between laptop computers or other portable devices, without a server, and even the exchange of floppy disk as the means of data transport."
The report was published online in the peer-reviewed Journal of Proteome Research and is expected to appear in print in early October. The article itself is available from IBM in PDF form.