0.24a + 0.41b + 0.35c + 0.12d + . . . = 1.00
0.21a + 0.29b + 0.20c + 0.08d + . . . = 1.00
0.41a + 0.23b + 0.22c + 0.05d + . . . well, you get the idea.
Thousands of equations like these - none of them particularly daunting in itself, save for the fact that it would be insoluble without the rest - together add up to a special kind of blood test. The bottom line is a newfound ability to figure out - across a broad diagnostic landscape - which important genes in which of several important cell types sitting in a blood sample have undergone an ominous change in their activity levels. And all this without having to physically separate the sample into its constituent cell types.
Just don't ask me how. I merely wrote the press release; I don't rate one.
Anyhow, using the new algorithm designed by post-doc Shai Shen-Orr, PhD, working in the labs of Mark Davis, PhD, and Atul Butte, MD, PhD, researchers were able to detect medically significant changes in blood that would otherwise go unnoticed. For instance, looking at whole blood drawn from 24 pediatric kidney-transplant patients, they could accurate assess which of those patients were showing early (i.e., pre-symptomatic) signs of organ rejection and which were accepting the grafts without complications.
Maybe not so far down the road, clinicians will be able to do this, too, as well as ferret out early warnings of cancer, autoimmune flare-ups and other diseases.
Blood draws are the coin of the realm. Second only to skin, blood is the most accessible tissue for analysis. And it's much more informative than skin, because it travels throughout the body, both acting on and reporting on distant and hidden tissues. But it's a bit messy because it contains so many different cell types, each doing its own thing.
"Any 7-year-old can look at a blood sample under a microscope and see it's a mix of a huge number of different kinds of cells," says Butte. But a change in gene-activity levels that, occurring in Cell Type A, might flag some news is a snooze if it's happening in Cell Type B.
Suppose a public-opinion analyst, new on the job, were to conduct two national voter-preference surveys before and after a politician's public speech, to see if that speech improved or impaired the politician's popularity among those who were present. But our rookie analyst has neglected to ask those surveyed which party they lean toward or what state they come from, so doesn't realize the first survey sample had a Democrat-to-Republican ratio of 30:70, while in the second, the ratio was reversed. The analyst might mistakenly infer a huge swing in pre- and post-speech voter sentiment, when in fact the only real change was in the samples' compositions. Meanwhile, a vehement change in support among residents of a small but election-swinging state might go undetected.
Likewise, medically significant changes in gene-expression patterns can go unnoticed in tests of an aggregate blood sample, while those that reflect only changes in the sample's composition can trigger false alarms.
Existing ways of physically telling cells apart either are pricey, time-consuming, and tedious, or don't lend themselves to microarray analysis - the workhorse technology used to analyze the activity levels of tens of thousands of genes at once. Shen-Orr, Davis, Butte, Stanford statistician Robert Tibshirani and their colleagues got around that. It's a bit like going around picking up discarded campaign buttons after the event, forcing everyone leaving the stadium afterward to show their drivers' license on the way out, and for good measure, so to speak, forcing a Breathalyzer test on them as they get in their cars - and being able, from these shards, to piece together the puzzle who each attendee is going to vote for come November.
Wait till the political consultants find out about this.