Microsoft vows to “solve cancer” in a decade. Hubris ensues.

If there’s one thing that irritates me more than government agencies making bold proclamations about making progress in cancer but not providing sufficient funding to have even a shot of realizing such ambitions (I’m talking to you, Cancer Moonshot), it’s people in other disciplines that are not cancer biology making bold proclamations about how they’re going to “solve” cancer or coming up with new “theories” to explain cancer. That’s not to say that cancer research can’t benefit from new perspectives from different sciences and disciplines can bring or new ways of thinking about the problem of cancer. I might seem arrogant, but, whether I am arrogant or not, I’m not that arrogant. What irritates me so much is that these scientists who are not cancer biologists inevitably come across as arrogantly overconfident, not to mention as condescending. The attitude seems to be: How come you cancer biologists never thought of this before? How come you never saw this before? Of course, in some cases, cancer biologists did think of this before and did see this before, but ended up rejecting it because it didn’t fit with the evidence.

Perhaps the best example of this occurred a few years ago when two astrophysicists, Paul Davies and Charley Lineweaver, decided to jump into the cancer research business with a concept they called atavism as a cause for cancer. Basically, the idea was that cancer is an evolutionary “throwback” to the dawn of intracellular life. Of course, having admittedly “no prior knowledge of cancer,” Lineweaver and Davies had stumbled upon a very old idea without realizing how old it was. Indeed, they seemed to think they were the first to have thought of it. As I pointed out at the time, there can be advantages to brining in scientists from different disciplines, but one consequence of doing so is that they often don’t know which hypotheses that have been considered before and rejected based on the evidence and therefore frequently act as though they were the first to have thought of a new hypothesis. As blogger Darren Saunders put it at the time, Lineweaver and Davies remind one of a doctor who reinvented calculus.

Or this:

Earlier this week, I sensed a similar, but related phenomenon when I started seeing headlines like this one in The Independent, Microsoft will ‘solve’ cancer within the next 10 years by treating it like a computer virus, says company. My first reaction when I read that headline was stunned disbelief that anyone could be so arrogantly ignorant as to make a statement that definitive without apparently knowing much about cancer—or biology for that matter. To be fair, I decided to read the article, because I know that headlines don’t always match what was actually said; let’s just say they tend to strip nuance from the statement.

Silly me:

Microsoft says it is going to “solve” cancer in the next 10 years.

The company is working at treating the disease like a computer virus, that invades and corrupts the body’s cells. Once it is able to do so, it will be able to monitor for them and even potentially reprogramme them to be healthy again, experts working for Microsoft have said.

The company has built a “biological computation” unit that says its ultimate aim is to make cells into living computers. As such, they could be programmed and reprogrammed to treat any diseases, such as cancer.


“The field of biology and the field of computation might seem like chalk and cheese,” Chris Bishop, head of Microsoft Research’s Cambridge-based lab, told Fast Company. “But the complex processes that happen in cells have some similarity to those that happen in a standard desktop computer.”

As such, those complex processes can potentially be understood by a desktop computer, too. And those same computers could be used to understand how cells behave and to treat them.

Yes, there is a resemblance between cancer and computing in much the same way that counting on your fingers resembles a supercomputer. The hubris of this project is unbelievably. Seriously> I thought antivaccinationists demonstrated the arrogance of ignorance, but they’ve got nothing on Microsoft. (Of course, it is Microsoft.) My reaction was virtually identical to Derek Lowe’s, only with more…Insolence. Indeed, he perfectly characterized the attitude of people like Linweaver, Davies, and now Bishop as a “Gosh darn it fellows, do I have to do everything myself?” attitude. Yes, those of us in cancer research and who take care of cancer patients do tend to get a bit…testy…when someone like Bishop waltzes onto the scene and proclaims to breathless headlines that he’s going to solve cancer in a decade because he has an insight that you stupid cancer biologists never thought of before: The cell is just a computer, and cancer is like a computer virus. (Hey, you know, viruses cause some cancers; so why not make the analogy to computer viruses?)

Basically, what Microsoft is doing is yet another machine learning approach to cancer. Don’t get me wrong. I don’t have any objection to computational approaches to biology, cancer, and the treatment of disease. Quite the contrary. What chaps my posterior here isn’t necessarily the concept. If you hose off the many layers of hubris and bullshit behind Microsoft’s initiative, there might be a germ of a good idea there. In fact, if you strip the bullshit away, you’ll see that even Microsoft seems to realize that it’s overpromising:

Microsoft says that solution could be with us within the next five or ten years.

Andrew Philips, who leads Microsoft’s biological computation group, told The Telegraph that in as little as five years it hopes to be able to develop a system for detecting problems. “It’s long term, but … I think it will be technically possible in five to ten years’ time to put in a smart molecular system that can detect disease.”

Um, I have news for you. There are lots of research groups who’ve been working on this sort of problem for a long time in clinical medicine and oncology. Indeed, check out this review article, which shows that, while there aren’t a huge number of scientific papers being published each year on machine learning tools to predict cancer and cancer recurrence, there are a respectable number, and that number is growing. Such tools are being applied to genomic and proteomic data—and have been for years. This is not a new thing. And notice what Andrew Phillips says: In five-to-ten years maybe he can come up with a smart molecular system to detect disease. Those of you who’ve read my many posts about overdiagnosis and overtreatment know that detecting cancer at ever earlier stages will not necessarily result in better outcomes or improved survival. It will, however, make overdiagnosis (i.e., the detection of subclinical disease that would never progress to cause a problem within the lifetime of the patient) much more likely, and overdiagnosis always leads to some degree of overtreatment. (See breast cancer and prostate cancer.) We’ve been down this road before.

Again, don’t get me wrong. Maybe Microsoft has a new way of applying machine learning to cancer. Maybe it has new ways of modeling the cellular processes that lead to cancer. If so, its software engineers would do well to talk less and code more, instead of saying something like this:

“The field of biology and the field of computation might seem like chalk and cheese,” Chris Bishop, head of Microsoft Research’s Cambridge-based lab, told Fast Company. “But the complex processes that happen in cells have some similarity to those that happen in a standard desktop computer.”

As such, those complex processes can potentially be understood by a desktop computer, too. And those same computers could be used to understand how cells behave and to treat them.

If that were possible, then those computers wouldn’t only be able to understand why cells behave as they do and when they might be about to become cancerous. They’d also be able to trigger a response within a cell, reversing its decision and reprogramming it so that it is healthy again.

Model intracellular processes leading to cancer and look for ways to reverse the process? Well, golly gee! Why didn’t cancer researchers think of that? It’s only what they’ve been trying to do for the last 100 years! What is systems biology but doing exactly that, using genomic, proteomic, and metabolomic data? What is “precision medicine” but almost exactly this? After over 15 years of having the tools to analyze the expression of every gene in a cell simultaneously and the computational power to model it, we’re only just scratching the surface of systems biology and computational biology, and Microsoft is going to “solve” this problem in five to ten years. Would that it were so easy! As another blogger put it, it “would be great if genetics were just one big Intel Core I7 that one could program in binary assembly language after decoding its instruction set, but I have doubts it’s that simple.”

Here’s the thing. Cancer biology like all biology, is probabilistic, not deterministic. Computers are deterministic. Their instructions consist of binary strings of 0s and 1s. True, computers can model probabilistic situations, the number of possible outcomes rapidly becomes incredibly large, and in cancer biology the number of potential interactions is astronomical. Worse, we don’t understand many of the alterations in cancer cells. As I’ve pointed out many times before, cancer cells are really messed up, and, worse, cancers themselves, thanks to the power of evolution, are made of a very heterogeneous bunch of cells with a very messed up genome. That’s why cancer researches like Derek Lowe (and I) get a bit testy reading this sort of thing:

I have beaten on this theme many times on the blog, so for those who haven’t heard me rant on the subject, let me refer you to this post and the links in it. Put shortly – and these sorts of stories tend to put actual oncology researchers in a pretty short mood – the cell/computer analogy is too facile to be useful. And that goes, with chocolate sprinkles on it, for all the subsidiary analogies, such as DNA/source code, disease/bug, etc. One one level, these things do sort of fit, but it’s not a level that you can get much use out of. DNA is much, much messier than any usable code ever written, and it’s messier on several different levels and in a lot of different ways. These (which include the complications of transcriptional regulation, post-transcriptional modification, epigenetic factors, repair mechanisms and mutation rates, and much, much, more), have no good analogies (especially when taken together) in coding. And these DNA-level concerns are only the beginning! That’s where you start working on an actual therapy; that’s what we call “Target ID”, and it’s way, way back in the process of finding a drug. So many complications await you after that – you can easily spend your entire working life on them, and many of us have.

And I haven’t even mentioned the role of processes like epigenetics, the immune system, and all the other myriad biological processes that contribute to cancer. Nor have I mentioned that using machine learning on the medical literature, as also proposed by Microsoft, will be limited by the fact that there are a lot of crappy studies in the literature. Then there’s the consideration that the analogy itself is suspect. Computers are designed, programmed, and debugged by human beings. Organisms and cancers are the result of millions of years of biological evolution.

I’ll leave Microsoft with this analogy, quoting Douglas Adams in The Hitchhikers’ Guide to the Galaxy, “Space is big. You just won’t believe how vastly, hugely, mind- bogglingly big it is. I mean, you may think it’s a long way down the road to the chemist’s, but that’s just peanuts to space.” Well, cancer is complicated. Microsoft will find out how vastly, hugely, mindbogglingly complicated it is. I mean, you might think it’s complicated to trick people into upgrading to Windows 10, but that’s peanuts compared to cancer.

An analogy, and a relevant xkcd cartoon:

Yep that about sums it up.