Team:Evry/Policy and Practices/Philosophy
From 2014.igem.org
(6 intermediate revisions not shown) | |||
Line 29: | Line 29: | ||
<font color="blue"> | <font color="blue"> | ||
<div align="center"><br><blockquote><i>“Synthetic biology is the engineering of biology”</i> | <div align="center"><br><blockquote><i>“Synthetic biology is the engineering of biology”</i> | ||
- | <br><br><div align="right"> | + | <br><br><div align="right">Képès, 2010</div></blockquote> |
</div></font> | </div></font> | ||
Line 81: | Line 81: | ||
But not all synthetic biologists admit that the field is – or should be – as rational and predictable as it ambitions to be. The behavior of biological systems are so complex, so variable, that it rather seems impossible ; and during our iGEM experience, we found many times that even with standardized parts and rigorous protocols, the biological systems often don't behave as expected. To overcome this, we often had to change our protocols in ways that we couldn't really explain, except by saying "it works better". And we do not think that this is specific to our team, nor to the iGEM competition. We believe that the whole field of synthetic biology as it is now truly cannot be compared to engineering in terms of rational and predictable design. | But not all synthetic biologists admit that the field is – or should be – as rational and predictable as it ambitions to be. The behavior of biological systems are so complex, so variable, that it rather seems impossible ; and during our iGEM experience, we found many times that even with standardized parts and rigorous protocols, the biological systems often don't behave as expected. To overcome this, we often had to change our protocols in ways that we couldn't really explain, except by saying "it works better". And we do not think that this is specific to our team, nor to the iGEM competition. We believe that the whole field of synthetic biology as it is now truly cannot be compared to engineering in terms of rational and predictable design. | ||
- | + | ||
Line 103: | Line 103: | ||
<font color="blue"><blockquote><div align="center"><i>"Unlike other engineering disciplines, synthetic biology has not developed to the point where there are scalable and reliable approaches to finding solutions. Instead, the emerging applications are most often kludges that work, but as individual special cases. They are solutions selected for being fast and cheap and, as a result, they are only somewhat in control."</i></div> | <font color="blue"><blockquote><div align="center"><i>"Unlike other engineering disciplines, synthetic biology has not developed to the point where there are scalable and reliable approaches to finding solutions. Instead, the emerging applications are most often kludges that work, but as individual special cases. They are solutions selected for being fast and cheap and, as a result, they are only somewhat in control."</i></div> | ||
- | <div align="right">Arkin and Fletcher</div></blockquote></font> | + | <div align="right">Arkin and Fletcher, 2006</div></blockquote></font> |
Synthetic biologists are often tempted to use kludges, because they work where rational design fail for unexplainable reasons. For example, if they observe that it's easier to transform a bacteria by electroporation than by conjugation, they will chose to use the former, even if their knowledge led them to believe that conjugation was more likely to work. If they notice that two parts put together give an unpredicted function to an organism and if this function is useful, they will probably continue to combine these two parts to produce the emergent function even if they can't explain why it appears. And if they don't have the result they want while following the rational protocol that should lead to this result, they often try again with different parameters and methods until it works, and then adapt their protocol and model accordingly. | Synthetic biologists are often tempted to use kludges, because they work where rational design fail for unexplainable reasons. For example, if they observe that it's easier to transform a bacteria by electroporation than by conjugation, they will chose to use the former, even if their knowledge led them to believe that conjugation was more likely to work. If they notice that two parts put together give an unpredicted function to an organism and if this function is useful, they will probably continue to combine these two parts to produce the emergent function even if they can't explain why it appears. And if they don't have the result they want while following the rational protocol that should lead to this result, they often try again with different parameters and methods until it works, and then adapt their protocol and model accordingly. | ||
Line 109: | Line 109: | ||
But if synthetic biologists really do behave like this, as depicted by O'Malley in "Making knowledge in synthetic biology. Designs meets kludge" (2009), then it would mean that they are doing the exact opposite of what they ambition and claim to do. Instead of a rational assembly of well defined and modular parts, synthetic biology would be made of kludgy designs. | But if synthetic biologists really do behave like this, as depicted by O'Malley in "Making knowledge in synthetic biology. Designs meets kludge" (2009), then it would mean that they are doing the exact opposite of what they ambition and claim to do. Instead of a rational assembly of well defined and modular parts, synthetic biology would be made of kludgy designs. | ||
- | Does that mean that this method is wrong? For some authors like Andrianantoandro | + | Does that mean that this method is wrong? For some authors like Andrianantoandro, who admit that synthetic biology so far has been rather kludgy, the irrational aspect of synthetic biologists' methods and results can be, and has to be overcome. They think it will become possible to achieve rational, engineer-like design in the future, thanks to the increasing standardization of the field. But other authors like O'Malley think on the contrary that not only kludges are unavoidable in biology, but also they are, actually, a better, more efficient method to produce results and knowledge, than the too rigorous method of the engineers. |
<br><div align="center"><font color="blue"><h4>Which method should be favored?</h4></font></div> | <br><div align="center"><font color="blue"><h4>Which method should be favored?</h4></font></div> | ||
+ | |||
+ | Synthetic biologists have always been aware that the complexity of living organisms would be an important obstacle to the construction of new parts and biological systems, and to the modification of existent systems. Among the properties of biological systems that interfere the most with mathematical predictions, we can for example mention the transcriptional noise, since there is an irreducible variability in gene expression, or mutations, which can appear spontaneously and randomly in any living organism. In fact, one of the most essential properties of life is the unpredictability of its behavior: for two exact same stimuli, in the same environment, a living organism's reactions can be significantly different. | ||
+ | |||
+ | But it is in full awareness of this essential unpredictability of biological organisms, and in spite of it, that synthetic biologists claim to be able to predict the behavior of simple biological systems, and to become the engineers of biology. This confidence comes from the fact that the development of systems biology has considerably increased our comprehension of the living systems as a whole, in all their complexity. System biologists use models and mathematical tools to describe and quantify the behavior of biological systems. They showed that complex behaviors in cells, that previously seemed emergent or random, could in fact be described by a set of several differential equations. This was taken by many biologists as proof that at least part of the biological systems could be explained in a logical and predictable ways, and hence that rational designs, equivalent to those used by engineers in mechanics, electronics, aeronautics, etc., could also be applied in biology. And even though the methods used until now may not have been as rational as synthetic biologists had hoped, many believe that the field can become more rational. | ||
+ | |||
+ | Andrianantoandro, for example, thinks that the noise due to the variability of cells behaviors can be overcome if instead of engineering single cells, we engineered cell populations. Thanks to cell-cell communication, the behavior of each cell would be synchronized, and the population would then become a reliable module with little noise. Other similar solutions keep arising in the field to overcome various unpredictable behaviors. | ||
+ | |||
+ | Some recent progress in synthetic biology are another reason to hope that the practice of the field could become more rigorous in the near future. The number of standardized parts in the Registry increased considerably in the last years. Some important and complex realizations have been achieved, like the synthesis of a precursor of the artemisinin by a yeast. And overall, the knowledge of how to build parts and how to assemble them has greatly improved in the last ten years, thanks to the experience synthetic biologists are accumulating, and thanks to the fact that all the results and parts are share in open source. | ||
+ | |||
+ | These are strong arguments in favor of the idea that synthetic biology can overcome kludges. For Serrano, it is essentially because the field is very young that so far, most of the circuits designed by synthetic biologists were made without standardized parts, and with a lot of tinkering. But with the increased standardization of parts and protocols, synthetic biology could maybe indeed become more like engineering in the future. | ||
+ | |||
+ | |||
+ | However, it is not certain that kludge in synthetic biology should be overcome. | ||
+ | Some authors believe, on the contrary, that this method provides efficient solutions, and allows unpredicted discoveries, and as such should be kept as a good method to produce both results and knowledge in synthetic biology. | ||
+ | |||
+ | <font color="blue"><blockquote><div align="center"><i>"Kludging should not be interpreted as a failure of synthetic biology, but as a highly creative and effective process."</i></div> | ||
+ | <div align="right">O'Malley, 2010</div></blockquote></font> | ||
+ | |||
+ | It is indeed easy to illustrate the fact that by tinkering, we can find solutions that may not have arisen with a too rigorous method. When a scientist makes models, they expect the biological system to behave as described mathematically. But when they discover that the experimental result significantly differs from the prediction, they try another model, then another, until the model can effectively describe the experimentally observed behavior of the system. This method could be qualified as trial and error. And the interesting point is that what eventually makes the scientist say that one model is better than the others, cannot be explained a priori. The main, and sometimes only argument in favor of choosing one model over the others, is that the scientist saw a posteriori that it worked. | ||
+ | It is common to find papers where the researchers propose a model, and give the values of a set of parameters for which the model works. But there is no explanation of how they found those values. Indeed, they didn't choose them for rational, explainable reasons, but by trial and error, they eventually found this set of values for which the system worked. | ||
+ | This justification could be seen as not sufficient to be scientific, because it only relies on experimental observations, and not on a theory that can be made and supported rationally independently from the experience. But in fact, whether this method meets the criteria for being a good and scientific method or not, it allows to effectively find the parameters for which the model works. Without trial and error, the researchers may never be able to find those parameters, so in a way it could be seen as a productive method. | ||
+ | |||
+ | A problem could arise, though. If the model only works with parameters that were determined experimentally without any other justification, it may be that the model is actually false, and that it was only by chance that a certain set of parameters made it looked like it accurately described the biological system. Because of this, it is important to try to find a rational explanation for the data that were only found by trial and error, in order to ascertain the accuracy of both the system and the data. | ||
+ | |||
+ | Using such inelegant and efficient methods can thus be very productive: it allows to test different hypothesis, to find new parameters, and to determine certain values. This is what Max Delbrück has described as being flexible and responsive to the variability of the system of study, in order not to prevent new discoveries. Whereas in comparison, a too rigorous approach can only produce expected results, without any room for novel insights. | ||
+ | |||
+ | Moreover, with a method where the result is not necessarily predictable nor stable over time, it it possible to build biological systems that evolve. Evolution in synthetic systems is something the scientists often try to avoid, because it would allow the system to lose its implemented function, and thus its utility. And it would also permit the system to gain via evolution new functions that would be detrimental for any reason. But if we leave the possibility for the systems to evolve and to not behave in different ways than those predicted, then they could potentially gain new interesting new functions. These functions could for example be profitable in an industrial point of view. Or we could gain, from the evolution of our engineered devices, a better understanding of the mechanisms of evolution. | ||
+ | |||
+ | In her paper, O'Malley takes into account all these advantages that derive from kludges, and concludes that this inelegant method should, in fact, not be avoided at all cost in synthetic biology. Even though being able to rationally control and build biological systems as engineers of biology is an enticing ambition, kludges should also be kept to produce efficient results and new knowledge. | ||
+ | |||
+ | |||
+ | <div align="center"><h4><font color="blue"> Conclusion </font></h4></div> | ||
+ | |||
+ | |||
+ | This study allowed us to confront one of the main ambition of synthetic biology to its actual practice. Most researchers in the field think that the construction of synthetic biological systems that would be controlled by human will is possible, if they can standardize the different constituents of living organisms and find rigorous methods to assemble them. Thanks to the contribution of systemic biology, and to the methods borrowed to engineering, synthetic biologists claim to be able to master in the future the irregular and unpredictable behavior of simple living organisms. They want to design biological systems that would follow a precise quantitative model, in a foreseeable and stable way. | ||
+ | |||
+ | However, if there are indeed more and more biological parts that are standardized and characterized, their behavior is not always the expected one, and often cannot be rationally explained. As a result, the way synthetic biology is currently practiced is not as rigorous as scientists would like it to be. This apparent failure is counterbalanced though by the fact that synthetic biologists still manage to build systems that work with kludge, and thus that a trial and error method should probably kept as a useful tool in the field, even if synthetic biologists were ever to truly become engineers of biology. | ||
+ | |||
+ | But since the complex interactions that occur in living organisms are increasingly well understood, and that the techniques to build them are more and more efficient, we can still believe that synthetic biology will indeed become more rational in the future. There has in fact already been some interesting results in the field, which lead us to keep hoping it will bring biological systems under our rational understanding and designing abilities. | ||
+ | |||
+ | |||
+ | |||
+ | <div align="center"><h4><font color="blue"> Bibliography </font></h4></div> | ||
+ | |||
+ | ANDRIANANTOANDRO, E., BASU, et als., "Synthetic biology: new engineering rules for an emerging discipline". <i>Molecular Systems Biology</i>, 2006, 2, 2006.0028. | ||
+ | |||
+ | ARKIN, A. P., FLETCHER, D. A., "Fast, cheap and somewhat in control", <i>Genome Biol.</i>, 2006, 7(8):114. | ||
+ | |||
+ | CAMERON, D. E., Bashor, C. J., & Collins, J. J., "A brief history of synthetic biology". <i>Nat Rev Micro</i>, 2014, 12(5), 381–390. | ||
+ | |||
+ | CHURCH G., "From systems biology to synthetic biology", <i>Molecular Systems Biology</i>, Volume 1, Issue 1, 2005. | ||
+ | |||
+ | EUROPEAN COMMISSION 6TH FRAMEWORK PROGRAMME, "Synbiology: An Analysis of Synthetic Biology Research in Europe and North America. Final Report on Analysis of Synthetic Biology Sector", Septembre 2006. | ||
+ | |||
+ | HEINEMANN M., PANKE S., "Synthetic biology -‐ putting engineering into biology." Bioinformatics 2006, 22:2790-‐2799. | ||
+ | |||
+ | ISALAN, M., LEMERLE, C., et als., "Evolvability and hierarchy in rewired bacterial gene networks." Nature, 2008, 452(7189), 840–845. | ||
+ | |||
+ | KEPES, F., La biologie de synthèse: développements, potentialités et défis. <i>Réalités industrielles</i>, 2010, no 1, p. 8-14. | ||
+ | |||
+ | O'MALLEY M., "Making knowledge in synthetic biology. Designs meets kludge", <i>Biological Theory</i>, vol. 4, n°4, 2009, p. 378-389. | ||
+ | |||
+ | SERRANO, L., "Synthetic biology: promises and challenges". <i>Molecular Systems Biology</i>, 2007. | ||
+ | |||
<!-- inserer texte de la page ici --> | <!-- inserer texte de la page ici --> |
Latest revision as of 03:58, 18 October 2014
Policy and Practices - Philosophy
Design in Synthetic Biology: rationality versus kludge
Contents |
Introduction
The following philosophical reflexion originates from our observation that there is an important contrast between what synthetic biologists ambition to do — a rational, rigorous design with controlled and predicted results — and how they actually realize their biological systems — by trial and error, and tinkering. This method is not specific to synthetic biology, but is more visible and surprising in this field because many synthetic biologists precisely claim that synthetic biology is the engineering of biology.
Though every part of the Registry is supposed to be well characterized and modular, and though the transformation protocols look rigorous and rational, we often didn't obtain the expected result when we assembled the parts into our biological system. Of course, the reason could have been that we made mistakes in our experimentations; but though we acknowledge our fallibility, we believe that in fact this problem is shared by all synthetic biologists : biological systems often don't behave according to our expectations and models.
In order to have better results, we often used a method that was not the most rational nor elegant one, but which allowed us to solve or bypass the unpredictability of biological behaviors. And as we discovered by reading the literature, this method is thought to be very common amongst synthetic biologists by some authors, who called it "kludge".
Even though it wasn't directly linked to our Sponge Patrol project, we decided to have this reflexion on the contrast between the ambitions and the actual practice of synthetic biology, because we believe that it is very important for those who want to be part of this new scientific field to step back from the lab and the computers for a moment, and to take a critical look at this field.
Though scientists have many different definitions for what is synthetic biology, a recurrent element of these definitions is the analogy with engineering. This comparison is supported by the idea that synthetic biology allows to build standardized parts of living systems, whose functions and properties are know ; and that those parts can be assembled together in any system thanks to a rational and modular protocol. Synthetic biologists should also be able to make a model of the constructed organism and to predict its behavior. Eventually, building new living systems should become as easy for synthetic biologists as assembling non-biological parts in a machine is easy to engineers.
“Synthetic biology is the engineering of biology”Képès, 2010
However, living organisms are exceedingly complex, and at first sight it looks impossible to fully understand them ; and if we don't, how could we possibly have enough control over them to force them to behave like mechanical, predictable machines? This question is important, because it is precisely what we are trying to do in iGEM: to design and build a "Genetically Engineered Machine". We are, like most synthetic biologists, assuming (or at least hoping) that living systems can be designed to act in a calculated and useful way, like the machines of engineers do. However, in truth it is obviously not so. Everyday during our iGEM experience, we discovered how unpredictable living systems can be - and even though we sometimes managed to have the result we expected, the protocols to achieve such a result were hardly as rational as engineers protocols. In biology, we never fully understand the living systems we are working on, and as a consequence, the design process of our genetically engineered machine is more often a trial-and-error process than a fail-proof one.
This observation has led us to wonder about the importance of kludging and tinkering in synthetic biology. The word "kludge" stands for "klumsy, lame, ugly, dumb but good enough", and seems to describe quite accurately what we are actually doing in synthetic biology, when we use inelegant but successful solutions to solve a problem when the rational design doesn't bear the expected results. In fact, we were led to notice that this not-always rational, but very efficient way of doing research is common in many scientific fields; but we thought that it was even more significant in synthetic biology, because the researchers of this field are precisely claiming that their design is rational and systematic. The contrast between what synthetic biology want to do and what they are actually doing is, as a result, even more apparent.
Kludge : "klumsy, lame, ugly, dumb, but good enough"
Inelegant but successful solution to a problem.
Thus, as we are synthetic biologists ourselves and are confronted everyday to the actual practice of the field in the lab, we thought that it was very important to take a step back and have an objective and reflexive look at design in synthetic biology. The questions we want to address in the following reflexion are the following:
- How can synthetic biologists pretend to achieve rational and predictable design of living systems?
- Is design in synthetic biology rational, of is it the result of kludging?
- Which method should be favored?
How can synthetic biologists pretend to achieve rational and predictable design of living systems?
In 2006, a European Commission that gathered many experts in synthetic biology offered the following definition of the new field:
“Synthetic biology is the engineering of biology: the synthesis of complex, biologically based (or inspired) systems which display functions that do not exist in nature. This engineering perspective may be applied at all levels of the hierarchy of biological structures – from individual molecules to whole cells, tissues and organisms. In essence, synthetic biology will enable the design of ‘biological systems’ in a rational and systematic way.”High-level Expert Group European Commission, 2006
We can notice here that synthetic biologists are confident that such designs can be rational and systematic, which means that like the machines or the computer programs of the engineers, the designed synthetic systems must be composed of parts that are fully understood and characterized. Thus the researchers should be able to assemble those parts and obtain the expected result.
Let's take a quick look at the history of synthetic biology to better understand how such an ambition could arise in a field working on living systems, which are intrinsically complex and unpredictable. With the development of systemic biology in the last decades, it has been observed that there is an organized hierarchy in cellular networks, between functional modules. And from there, the idea appeared that some mechanisms in living organisms could be reduced to simple mechanisms such as those built by engineers in machines ; which meant that with the proper tools, we could modify and design cells in a rational way to obtain a predefined result. Then in 2004, the first international conference on synthetic biology, SB 1.0, organized by the MIT, brought together researchers specialized in fields as varied as biology, chemistry, physics, engineering and informatics. It is probably during this interdisciplinary meeting that the ambition to use the engineers' rational bottom-up approach in molecular biology and genetics took shape.
In order to achieve a rational design of living systems, the priority was first given to the characterization of their functional parts, which is the main goal of the iGEM competition. These researchers had the intuition that it would then become possible to assemble these parts in different ways, that don't necessarily exist in nature, using rational and systematic protocols ; and to construct biological devices that would act exactly as they were intended to.
"As envisioned by SynBERC, synthetic biology is perhaps best defined by some of its hallmark characteristics: predictable, off-the-shelf parts and devices with standard connections, robust biological chassis (such as yeast and E. coli) that readily accept those parts and devices, standards for assembling components into increasingly sophisticated and functional systems and open-source availability and development of parts, devices, and chassis.”SynBERC
Nowadays, we are able to build or synthesize several components of a simple cell, like the lipid bilayer or several enzymes. But the the element we use the most in synthetic biology, and in particular in iGEM, is the DNA (or the RNA). The reason for that is evident: most of the properties of living beings are encoded in their genome, so it is by modifying the genomes that we can effectively build organisms with new functions. But the ambition of synthetic biology is not just to synthesize existing genomes, nor even to just modify them ; its ambition is to fully characterize and standardize some genome sequences that code for a specific function. That is why at iGEM, we try to design plasmids called Biobricks, that contain a part which function is known, as well as specific sequences that allow the part to be cut and moved from the plasmid with simple protocols.
The genes we design in synthetic biology are often simplified compared to the average complexity of genes in living systems. Only four main elements are taken into account, because most of the time they are enough to express a gene in an organism: a promotor, a ribosome-binding sequence, a coding sequence and a terminator. And this choice also makes the designed constructions modular, which is a very important property in synthetic biology. A modular promotor for example is a promotor that can be moved from one gene to another, and keep the same properties in both constructions. Each gene can in fact be cut into small modular pieces ; or can be build out of small pieces of DNA - usually in plasmids, that are also modular and can replicate independently of the host genome, whatever the host species.
The sequences of all the Biobricks are stored in the Registry of Standard biological parts, and everyone can have access to them. This Registry is meant to be a catalogue of parts of the living machines: people can pick the parts they need, buy them, and then assemble them in a simple and rational way to build the devices of their choice - just like engineers can buy electronic components from catalogues and then assemble them to build machines.
The other main reason that is supposed to make design in synthetic biology rational and predictable, is the use of the engineers' methodology for the design and construction of biological devices: conception and modeling, construction, implementation and validation. In the first step, synthetic biologists use mathematics and informatics tools to make a model of the system and simulate its behavior in silico. The parts of the system are then build or synthesized, before being implemented into a machine or a living organism. The researchers then test whether the constructed biological system behave as predicted — and if it doesn't, they change the model and try again. By using this method, synthetic biologists hope to bring more rigor to the field.
But not all synthetic biologists admit that the field is – or should be – as rational and predictable as it ambitions to be. The behavior of biological systems are so complex, so variable, that it rather seems impossible ; and during our iGEM experience, we found many times that even with standardized parts and rigorous protocols, the biological systems often don't behave as expected. To overcome this, we often had to change our protocols in ways that we couldn't really explain, except by saying "it works better". And we do not think that this is specific to our team, nor to the iGEM competition. We believe that the whole field of synthetic biology as it is now truly cannot be compared to engineering in terms of rational and predictable design.
Is design in synthetic biology rational,
or is it the result of kludging?
In truth, synthetic biology is rarely as rational as it claims to be. The attempts to control biological systems continuously clash with the unpredictable behavior of living organisms. In the iGEM competition for example, we very often discover that a part from the Registry doesn't have the expected function when added to our biological device: its expression change depending on the chassis in which it is introduced, on its interaction with other components present in the cell, and on the media in which the cell grows. The problem with the so-called modular parts of the Registry is that it is always characterized in particular conditions; but its behavior can never be exactly the same in different conditions. In biological systems, the slightest modification of the environment can result in considerable changes in genetic expression. This is why the uniformity and exact reproducibility of a function coded by a genetic part in several conditions are still out of reach - and may be forever impossible - even if the part was correctly characterized in particular conditions.
"The behavior of [small] modules is affected in a large extent by the rest of the network in which they are embedded"Isalan et al., 2008
When we work with the parts of the Registry, we must also face the fact that many parts have not been well characterized. Out of the 5000 parts that we can buy from the Registry, only about 2000 have been verified by researchers others than those who have built them. Since an important number of the parts of the Registry have been designed by students, who only had a limited time to build and characterize them for the iGEM competition, many of these parts are ill-assembled or do not have the function described on the Registry.
Moreover, even if the function of several parts is well-known, they may not behave as expected when we assemble them together in the same plasmid or in the same organism. Unpredicted emergent properties can arise in such systems. In 2005, Church claimed that by combining biological parts together, we would be able to overcome the indeterminate and non modular aspect of living organisms, and to build biological systems with steady behaviors. But now, more and more biologists agree that if we combine together several well characterized parts, the new system may not only have the addition of the functions of each part: new properties and unpredicted behaviors can very well arise. Which is an obstacle to rational design. The more complex the system, the more likely it is that the obtained result will differ from the expected one. As a result, synthetic biology is more often a long process of trial and errors, than a rational and functional design.
As design often leads to unexpected results, it's in fact a more pragmatic approach that is often taken by synthetic biologists, especially at iGEM where we have to deal with the necessity to produce results in a very short amount of time: the method with the best results is a posteriori chosen over the method that looked a priori more rational. For Oliver Müller, the scientific design is often idealized by biologists, who claim to know exactly how to obtain a certain result, when in fact biological systems are very complex and beyond their control. The uncertainty and the unexpected, in experimental design, are often left untold in synthetic biology as well as in other scientific fields. The reason is that scientists need to frequently publish and show positive results. Negative results, in most cases, are not published. Moreover, research in biology is a competitive field, where scientists are often tempted to show that their results are the outcome of a rational, well-planned design, even when the results were not those predicted at all. In fact, we think that in iGEM for example, students often change their initial objectives when they present their project, to make them match the results they actually got ; this way, it looks like the project worked just as planned, when in fact it is the result of unexpected results and tinkering.
For those reasons, O'Malley and other authors have portrayed design in synthetic biology as a pragmatic tinkering, where researchers use parts without knowing if they will assemble correctly, nor what will be the final result. If in theory the synthetic biologist should be able to understand and explain each step of the design and the behavior of the system, in fact the priority is given to results and not to rational design, in order to be able to publish, or to have more funds, or in our case to win the iGEM competition. This method has been called "kludge" by those authors, which is a term that comes from computer science:
"Unlike other engineering disciplines, synthetic biology has not developed to the point where there are scalable and reliable approaches to finding solutions. Instead, the emerging applications are most often kludges that work, but as individual special cases. They are solutions selected for being fast and cheap and, as a result, they are only somewhat in control."Arkin and Fletcher, 2006
Synthetic biologists are often tempted to use kludges, because they work where rational design fail for unexplainable reasons. For example, if they observe that it's easier to transform a bacteria by electroporation than by conjugation, they will chose to use the former, even if their knowledge led them to believe that conjugation was more likely to work. If they notice that two parts put together give an unpredicted function to an organism and if this function is useful, they will probably continue to combine these two parts to produce the emergent function even if they can't explain why it appears. And if they don't have the result they want while following the rational protocol that should lead to this result, they often try again with different parameters and methods until it works, and then adapt their protocol and model accordingly.
But if synthetic biologists really do behave like this, as depicted by O'Malley in "Making knowledge in synthetic biology. Designs meets kludge" (2009), then it would mean that they are doing the exact opposite of what they ambition and claim to do. Instead of a rational assembly of well defined and modular parts, synthetic biology would be made of kludgy designs.
Does that mean that this method is wrong? For some authors like Andrianantoandro, who admit that synthetic biology so far has been rather kludgy, the irrational aspect of synthetic biologists' methods and results can be, and has to be overcome. They think it will become possible to achieve rational, engineer-like design in the future, thanks to the increasing standardization of the field. But other authors like O'Malley think on the contrary that not only kludges are unavoidable in biology, but also they are, actually, a better, more efficient method to produce results and knowledge, than the too rigorous method of the engineers.
Which method should be favored?
Synthetic biologists have always been aware that the complexity of living organisms would be an important obstacle to the construction of new parts and biological systems, and to the modification of existent systems. Among the properties of biological systems that interfere the most with mathematical predictions, we can for example mention the transcriptional noise, since there is an irreducible variability in gene expression, or mutations, which can appear spontaneously and randomly in any living organism. In fact, one of the most essential properties of life is the unpredictability of its behavior: for two exact same stimuli, in the same environment, a living organism's reactions can be significantly different.
But it is in full awareness of this essential unpredictability of biological organisms, and in spite of it, that synthetic biologists claim to be able to predict the behavior of simple biological systems, and to become the engineers of biology. This confidence comes from the fact that the development of systems biology has considerably increased our comprehension of the living systems as a whole, in all their complexity. System biologists use models and mathematical tools to describe and quantify the behavior of biological systems. They showed that complex behaviors in cells, that previously seemed emergent or random, could in fact be described by a set of several differential equations. This was taken by many biologists as proof that at least part of the biological systems could be explained in a logical and predictable ways, and hence that rational designs, equivalent to those used by engineers in mechanics, electronics, aeronautics, etc., could also be applied in biology. And even though the methods used until now may not have been as rational as synthetic biologists had hoped, many believe that the field can become more rational.
Andrianantoandro, for example, thinks that the noise due to the variability of cells behaviors can be overcome if instead of engineering single cells, we engineered cell populations. Thanks to cell-cell communication, the behavior of each cell would be synchronized, and the population would then become a reliable module with little noise. Other similar solutions keep arising in the field to overcome various unpredictable behaviors.
Some recent progress in synthetic biology are another reason to hope that the practice of the field could become more rigorous in the near future. The number of standardized parts in the Registry increased considerably in the last years. Some important and complex realizations have been achieved, like the synthesis of a precursor of the artemisinin by a yeast. And overall, the knowledge of how to build parts and how to assemble them has greatly improved in the last ten years, thanks to the experience synthetic biologists are accumulating, and thanks to the fact that all the results and parts are share in open source.
These are strong arguments in favor of the idea that synthetic biology can overcome kludges. For Serrano, it is essentially because the field is very young that so far, most of the circuits designed by synthetic biologists were made without standardized parts, and with a lot of tinkering. But with the increased standardization of parts and protocols, synthetic biology could maybe indeed become more like engineering in the future.
However, it is not certain that kludge in synthetic biology should be overcome.
Some authors believe, on the contrary, that this method provides efficient solutions, and allows unpredicted discoveries, and as such should be kept as a good method to produce both results and knowledge in synthetic biology.
"Kludging should not be interpreted as a failure of synthetic biology, but as a highly creative and effective process."O'Malley, 2010
It is indeed easy to illustrate the fact that by tinkering, we can find solutions that may not have arisen with a too rigorous method. When a scientist makes models, they expect the biological system to behave as described mathematically. But when they discover that the experimental result significantly differs from the prediction, they try another model, then another, until the model can effectively describe the experimentally observed behavior of the system. This method could be qualified as trial and error. And the interesting point is that what eventually makes the scientist say that one model is better than the others, cannot be explained a priori. The main, and sometimes only argument in favor of choosing one model over the others, is that the scientist saw a posteriori that it worked. It is common to find papers where the researchers propose a model, and give the values of a set of parameters for which the model works. But there is no explanation of how they found those values. Indeed, they didn't choose them for rational, explainable reasons, but by trial and error, they eventually found this set of values for which the system worked. This justification could be seen as not sufficient to be scientific, because it only relies on experimental observations, and not on a theory that can be made and supported rationally independently from the experience. But in fact, whether this method meets the criteria for being a good and scientific method or not, it allows to effectively find the parameters for which the model works. Without trial and error, the researchers may never be able to find those parameters, so in a way it could be seen as a productive method.
A problem could arise, though. If the model only works with parameters that were determined experimentally without any other justification, it may be that the model is actually false, and that it was only by chance that a certain set of parameters made it looked like it accurately described the biological system. Because of this, it is important to try to find a rational explanation for the data that were only found by trial and error, in order to ascertain the accuracy of both the system and the data.
Using such inelegant and efficient methods can thus be very productive: it allows to test different hypothesis, to find new parameters, and to determine certain values. This is what Max Delbrück has described as being flexible and responsive to the variability of the system of study, in order not to prevent new discoveries. Whereas in comparison, a too rigorous approach can only produce expected results, without any room for novel insights.
Moreover, with a method where the result is not necessarily predictable nor stable over time, it it possible to build biological systems that evolve. Evolution in synthetic systems is something the scientists often try to avoid, because it would allow the system to lose its implemented function, and thus its utility. And it would also permit the system to gain via evolution new functions that would be detrimental for any reason. But if we leave the possibility for the systems to evolve and to not behave in different ways than those predicted, then they could potentially gain new interesting new functions. These functions could for example be profitable in an industrial point of view. Or we could gain, from the evolution of our engineered devices, a better understanding of the mechanisms of evolution.
In her paper, O'Malley takes into account all these advantages that derive from kludges, and concludes that this inelegant method should, in fact, not be avoided at all cost in synthetic biology. Even though being able to rationally control and build biological systems as engineers of biology is an enticing ambition, kludges should also be kept to produce efficient results and new knowledge.
Conclusion
This study allowed us to confront one of the main ambition of synthetic biology to its actual practice. Most researchers in the field think that the construction of synthetic biological systems that would be controlled by human will is possible, if they can standardize the different constituents of living organisms and find rigorous methods to assemble them. Thanks to the contribution of systemic biology, and to the methods borrowed to engineering, synthetic biologists claim to be able to master in the future the irregular and unpredictable behavior of simple living organisms. They want to design biological systems that would follow a precise quantitative model, in a foreseeable and stable way.
However, if there are indeed more and more biological parts that are standardized and characterized, their behavior is not always the expected one, and often cannot be rationally explained. As a result, the way synthetic biology is currently practiced is not as rigorous as scientists would like it to be. This apparent failure is counterbalanced though by the fact that synthetic biologists still manage to build systems that work with kludge, and thus that a trial and error method should probably kept as a useful tool in the field, even if synthetic biologists were ever to truly become engineers of biology.
But since the complex interactions that occur in living organisms are increasingly well understood, and that the techniques to build them are more and more efficient, we can still believe that synthetic biology will indeed become more rational in the future. There has in fact already been some interesting results in the field, which lead us to keep hoping it will bring biological systems under our rational understanding and designing abilities.
Bibliography
ANDRIANANTOANDRO, E., BASU, et als., "Synthetic biology: new engineering rules for an emerging discipline". Molecular Systems Biology, 2006, 2, 2006.0028.
ARKIN, A. P., FLETCHER, D. A., "Fast, cheap and somewhat in control", Genome Biol., 2006, 7(8):114.
CAMERON, D. E., Bashor, C. J., & Collins, J. J., "A brief history of synthetic biology". Nat Rev Micro, 2014, 12(5), 381–390.
CHURCH G., "From systems biology to synthetic biology", Molecular Systems Biology, Volume 1, Issue 1, 2005.
EUROPEAN COMMISSION 6TH FRAMEWORK PROGRAMME, "Synbiology: An Analysis of Synthetic Biology Research in Europe and North America. Final Report on Analysis of Synthetic Biology Sector", Septembre 2006.
HEINEMANN M., PANKE S., "Synthetic biology -‐ putting engineering into biology." Bioinformatics 2006, 22:2790-‐2799.
ISALAN, M., LEMERLE, C., et als., "Evolvability and hierarchy in rewired bacterial gene networks." Nature, 2008, 452(7189), 840–845.
KEPES, F., La biologie de synthèse: développements, potentialités et défis. Réalités industrielles, 2010, no 1, p. 8-14.
O'MALLEY M., "Making knowledge in synthetic biology. Designs meets kludge", Biological Theory, vol. 4, n°4, 2009, p. 378-389.
SERRANO, L., "Synthetic biology: promises and challenges". Molecular Systems Biology, 2007.