Biology is in a slump - we need error logs to get out of it
It isn’t lost on me that what I am about to say is rather cliche and pronounced at least twice a month somewhere in the world. I feel that biological research isn’t progressing like it should be. First some background.
The two paragraph background
In circa 2000, about the time I was looking at grad schools, I read the book “Gene Dreams.” The book rehashes the biotech scene of the 1980’s and 90’s, along with all of the promises that biotech + gene therapy would bring to the world. It notes that all of those ‘dreams’ failed, but that the 2000’s would make it right. I was a believer! I entered graduate school determined to use gene and stem cell therapy to cure the entire F'ing planet. Spoiler alert, I failed.
I wasn’t alone though. Many others have failed in that pursuit as well. Gene and stem cell therapy is largely an 'applied science’ versus being a 'basic science’ (aka learning about the building blocks of what makes things tick). So, for a long time I felt that it was just applied science that had issues. The “issues” being that no solid or common frameworks to make discoveries or breakthroughs exist. And that it was better to do basic science if you wanted to succeed more easily. Years after completing grad school though, I think my bitterness was too narrow. I’m an equal opportunist, and would therefore like to bash basic science as well for lacking a modern toolset to take “it” (whatever “it” is) to the next level.
That’s the quick, slightly incoherent two paragraph background, now on to the (somewhat) present thinking.
I was at FooCamp last June 2012 and wound up in a session about biology (don’t remember the title). I think a discussion was brought up originally over the question of how programmers could get more involved with science in a DIY approach by Jesse Robbins. By this time, I had been several years removed from doing actual science on an actual bench with an actual pipette. Instead, I had taken up what I originally started out doing (at age 5) by programming shit; albeit shit related to science now.
Something that was said during that discussion brought me back to my graduate school days of failed experiment after failed experiment. Jesse asked, “Jason, how can I get more involved in science?”
My first reaction was, “Don’t! You’ll end up depressed.” All of those failed experiments, and all of the failed work by others in gene therapy, and most of the time we had no-f-ing-clue-why-they-were-failing. That was the biggest drain on me in grad school, not knowing why. Equally, if I did some kind of PCR directed-evolution experiment and found ONE good result amongst a thousand failed wells in a plate, my next step would be to toss those 999 failed wells and evolve the good well in another PCR round. How, stupid, was, I.
Having gotten back into programming after grad school, I immediately realized my error. Instead of concentrating on the one good well in a thousand, I should have taken the 999 bad wells to see why they went wrong. In programming, this is debugging by looking at your error logs, or similarly, test-driven development.
Moreover, what was missing was some sort of stack trace that would tell me exactly at which point the antibiotic I was testing started to kill the cell all the way up to the point of death (antibiotics were often used to test PCR evolved enzymes used in gene/cell therapy). It was then that I realized that not just applied science, but basic science was missing these “error logging” tools as well. We usually don’t know why experiments fail because 1) we hardly care about failures in science and 2) no one, AFAIK, has built the tools to examine the “why” stuff fails. Such an approach would be laughed at in programming.
Instead, in programming you check your error logs to figure out how to make things work as expected. And the reason we can do that is because someone in the past spent the time building error reporting tools. What is missing in biology (again, AFAIK) is a comprehensive set of error logging within cells.
This is why we’re in a progression slump; we can’t make the huge mental leaps in science because we’re now at the point where better tools are needed. This happens once a generation or so, where we’ve surpassed the infrastructure that was built by the previous generation. A more methodical approach within a framework is needed. A new respect for failed experiments is needed.
And I wouldn’t be surprised if building such error reporting within cells was more universal than we might expect at first, no matter what the cell type or species whence it came. Error reporting I suspect, much like DNA itself across species, could be generalized enough to require little modification per cell, protein interaction, or whatever. A common set of error messages could bubble up to a logging system to be investigated post-experiment. Even the act of building such error reporting would result in a giant number of new discoveries.
Of course, like any idea, good or bad, this has probably been “thunk” before, possibly at the same time. The thing is, I’ve yet to see anyone do it. Perhaps because they did and it failed, and so the research was tossed per usual.
