By Christina Agapakis | October 15th 2009 10:30 AM | Print | E-mail

    Failure is enjoying something of a resurgence in pop culture. Blogs devoted entirely to failures of "epic" proportions have made it somehow appropriate to scream "FAIL" at people who have already been publicly humiliated, while self-help columns, university commencement speeches, and business advice books focus on the importance of failure in developing a stronger personality, growing up, or starting up a successful business. Failure is everywhere, and now it plays an important role in our internet-based entertainment as well as our personal, societal, and economic growth. Despite changes in trendiness, failure has always been an integral part of how we learn from life, and how we learn about life in the study of biology.

    As a graduate student, I am well acquainted with failure. Like Inuit words for snow, graduate students don't have any more words for failure than everyone else, but we do experience many different types of failure, failures in classwork, failure of experiments, failure of hypotheses. There's failure you can blame on your labmates, on lab supply companies, on God, on lack of sleep. There's the magical failure that you can learn from and become a better scientist (until your third year) and there's the failure that convinces you that there must be some kind of karmic retribution going on for all those helpless E. coli you killed in the course of your experiments.

    Complaining about the daily failures of graduate work is of course better suited for department beer hours than anything in writing, and we don't learn much from these failures besides how many experiments one person can reasonably do at once and why not to run gels backwards. But, like those self-help books say, we can look at some of the larger-scale failures--of common hypotheses, entrenched theories, and scientific paradigms--in order to better understand the processes and production of scientific knowledge, growing as scientists. 

    Moreover, studying these historical failures, one sees that the notion of what constitutes a failed experiment is something that can change over time, slowly altering the process of academic science and scientific publishing with it. Recently, unexpected or negative results previously hidden away in a shamed lab notebook have begun to see the light of day thanks to publications like the Journal of Negative Results in Biomedicine, as well as some of the more casual routes of science communication allowed by the internet. This "dark data" may prove useful for the work of other researchers, or at least help prevent someone from following the same flawed hypothesis further down the line. Graduate students may always have to deal with the everyday experimental failures, but understanding the larger ones in context can help all scientists do better work. 

    Even when looking at the larger context of failure in academic science, this kind of talk about failure still sounds mostly like the usual platitudes about "learning from your mistakes" and "if at first you don't succeed…" How does failure affect how we actually learn about biology? What role does it play in those elusive successful experiments? Scientific successes in the life sciences, the development of new life-saving drugs or therapies, new understandings into the deep workings of the cell, these successes are built on a rich history of science, a history of failed experiments, entire failed theories or even failed fields of study. However, the most successful results in biology come from experiments into how biology itself fails. Today's biology is built on what scientists were able to learn by studying diseases and injuries, mapping mutations, and deleting whole chunks of genomes to see what happens. 

    Genes or systems without a clear failure mode, an observable change in appearance or behavior when they break, are much less well understood than those that can be studied through deletion or mutation. There are more yeast researchers than there are yeast genes, but there are still many uncharacterized genes in the yeast genome, often because deleting these genes doesn't seem to harm the organism in any way. By systematically deleting genes in model organisms and screening for desired behaviors, genetics has been able to identify genes that contribute to a tremendous number of cellular processes, systems that can often be generalized to many other organisms, including humans. Genetic screens seeking cells that couldn't divide after having their genomes mutated identified many of the genes that control cellular growth and mitosis, genes that play an important role in many human cancers. Genes that control the lifespan of nematode worms, and perhaps may play a role in human aging, were discovered by randomly damaging parts of the worm genome and looking for individuals that lived for longer or shorter times than their wild-type cousins. 

    Even in our own bodies, some of the best characterized genes and systems are those that are involved in a disease state, especially those diseases that have a clear genetic component. Almost everything we know about cholesterol in our bodies comes from studying the genes and cells of individuals with a rare form of familial high cholesterol. In families carrying a certain genetic mutation, almost all family members have extraordinarily high cholesterol from a young age. By teasing apart the details of the proteins that were affected by this mutation, scientists were able to understand much of the cellular machinery involved in processing cholesterol from the blood. At the scale of whole organs, our understanding of the anatomy of the human brain is to a large extent defined by observing the behavior of people with tumors or other lesions in specific brain regions, as popularized in Oliver Sacks' book, The Man Who Mistook His Wife for a Hat. The eponymous Man helped Dr. Sacks understand how the brain processes visual information. While his eyes could see perfectly, a lesion in his visual cortex prevented his brain from processing the information, from understanding shapes and colors he saw as belonging to a hat or a person. These rare cases of failure of human circuitry allow us to better understand how healthy brains work, taking the shapes the eyes see and turning them into objects with meaning.

    How does this focus on disease affect us as scientists and as people? It is possible that a fuller understanding of human physiology based on health as well as disease may help us to live healthier lives through a preventative medicine. Beyond individual health, the notion of mutation, central to our understanding of evolution at the scale of organisms, cells, and molecules is in our language essentially a negative word. "Mutant" brings to mind many gruesome horror movies, but without mutation, there is no evolution, there is no life. In a sense, we are all mutants. Each person has a slightly different genetic code, and no one can decide which sequence is the standard by which to compare all others. As a driver of life, change, and diversity, mutation should be considered much more than simple molecular mistakes, failures of the replication or recombination machinery.

    Alongside these analytic, top-down approaches that break down life in order to understand it, synthetic, bottom-up methods have tried to recreate living processes, to understand living things by constructing them.  First coined in the early 20th century by Stéphane Leduc, the term synthetic biology was originally used to refer to research on physical processes that spontaneously resembled living cells. Since then, synthetic biology has developed in parallel with our increased understanding of the molecular details of biological systems. From the motions of colored chemical salts in water, to the artificial transfer of genetic material between organisms, to the engineering of cellular switches and logic gates, to building entire genomes in vitro, synthetic biology has meant many things but primarily represents a parallel constructive paradigm for the study of biology.

    Synthetic biology emerged a few decades after synthetic chemistry, and while it has made tremendous progress has yet to reach the level of sophistication that exists in chemical synthesis. In chemistry, synthesis allowed for a deeper understanding of how atoms connect to form molecules, something impossible to do when simply breaking down molecules to identify their atomic constituents. In biology, synthetic approaches may have a similar impact on how we understand the way that biological molecules connect to make living biological processes. By attempting to design and construct networks of genes, recreating living processes starting with just DNA, RNA, lipids, and proteins (and usually cells that are already alive), it may be possible to understand how the connections between these chemical molecules lead to the emergent behaviors that characterize living things: how genomes become life.

    This constructive approach has many important implications for ethics, politics, biosafety, and importantly, the study of biology. Failure is no longer the start of a scientific discovery, and understanding failed synthetic experiments becomes much more complicated. A synthetic biological system that works implies that our picture of how the components work together is correct. A synthetic system that fails shows you that you need to do more research. Often, failure exists in parallel with success, as seen with an early example of the more recent incarnation of engineering-based synthetic biology--the repressilator. This biological system is built out of a ring of three genes where each gene represses the expression of the next. By tightly controlling the strength of the expression and the repression, the system will create an oscillation in the expression of a fourth, glowing gene. The synthetic system works, oscillates, starting dim, then glowing green, then turning off again, showing that the model the authors used of how genes can connect to create oscillatory behavior was predictive. However, it only works for a very short time, going through just a few bright and dark cycles before shutting off for good, and the cells aren't synchronized with one another, each glowing and turning off at different times. As a model of how oscillations like brain waves or heart beats or circadian rhythms function, where many cells work in conjunction and maintain an oscillation for years, the repressilator is far off. 

    Synthetic biological systems can build off of previous engineered pathways just as hypotheses and theories in other types of research rely on one another, and additional machinery can be engineered in to try and improve the original designs. Recent additions to the repressilator system have allowed cells to communicate and oscillate in synchrony, and other additions have helped to integrate the oscillations more closely with the natural cellular machinery. However, not all failures can point to a clear path forward. When a synthetic system doesn't behave in the predicted way, doesn't produce any behavior, or worse, kills the cell, these failures can imply any number of possible explanations, from the mundane experimental problems of everyday labwork, to an incomplete understanding of the system in question. The ambiguity of failures in synthetic biology can be seen as problems to be solved by engineers, but maybe they're just part of life.


    Chand, Sudeep. "We're all mutants, say scientists." BBC News, 2009. <>

    Elowitz, Michael B. and Stanislas Leibler. "A synthetic oscillatory network of transcriptional regulators." 2000, Nature, 403(6767): 335-338.

    Goetz, Thomas. "Freeing the Dark Data of Failed Scientific Experiments." Wired Magazine, 15.10., 2007. <>

    Keller, Evelyn Fox. "Making Sense of Life." 2002, Harvard University Press, Cambridge, MA.

    Sacks, Oliver. "The Man Who Mistook His Wife For a Hat." 1985, Summit Books, New York, NY.

    Yeh, Brian J. and Wendell A. Lim. "Synthetic biology: lessons from the history of synthetic organic chemistry." 2007, Nature Chemical Biology, 3(9): 521-529.