Is It Becoming Acceptable to Speak of Biological Systems and Processes in Terms of Design?
To the question posed in the headline, the answer is: It seems that way sometimes. And can speaking about design in such a context be done without getting hammered by the press, censored, or ridiculed? Perhaps. We’ll see. In the following example, think of the Darwinese as packing peanuts that can be removed to get to the important items inside.
A remarkable paper was published in BioEssays in January, with three authors from the University of Washington, Steven S. Andrews, H. Steven Wiley, and Herbert M. Sauro. None has any known sympathies for intelligent design. And yet much of their paper, “Design patterns of biological cells,” could have been written by any one of the PhDs presenting ideas at the Conference on Engineering in Living Systems (CELS).
Design patterns are generalized solutions to frequently recurring problems. They were initially developed by architects and computer scientists to create a higher level of abstraction for their designs. Here, we extend these concepts to cell biology to lend a new perspective on the evolved designs of cells’ underlying reaction networks. We present a catalog of 21 design patterns divided into three categories: creational patterns describe processes that build the cell, structural patterns describe the layouts of reaction networks, and behavioral patterns describe reaction network function. Applying this pattern language to the E. coli central metabolic reaction network, the yeast pheromone response signaling network, and other examples lends new insights into these systems.
Taken for Granted
The authors do not question Darwinian evolution, taking it for granted some 14 times in the paper. They speak of “the evolution of complex life” and convergent evolution, even speculating on whether life on other planets would evolve the same way as it has on Earth. Such talk is common in biomimetics literature as well: e.g., one writer spoke of an ingenious solution that was “refined over more than 420 million years of evolution,” as if natural selection gave an organism a head start. We can safely dismiss such statements as either poetic license or a misunderstanding of evolution in its usual unguided sense.
The important items are these: a catalog of 21 design patterns presented as solutions to engineering problems that cells have solved. Here’s one example:
Pores and pumps
Problem
Cellular components, from ions to proteins, typically need to be localized to the correct sides of membranes, including the plasma membrane, nuclear membrane, and other organelle membranes.
Solution.
Trans-membrane pores and pumps that use either active or passive transport. These pores and pumps are typically quite selective about what molecules they transmit and are often gated by external signals.
Cell membranes are quite permeable to oxygen, carbon dioxide, and other small nonpolar molecules but are effectively impermeable to larger and more charged species, a property that is essential to establishing and maintaining cell organization. Transport of these latter species occurs via transporters and channels, including ion channels, passive and active transporters for ions or other small molecules, proton pumps, ABC transporters, photosynthetic reaction centers for electron transport, and ATP synthase proteins for mitochondrial proton transport. The nuclear pore complex is a particularly large pore, which enables passive transport of small molecules and performs active transport on proteins that carry nuclear localization or nuclear export signals.
Readers can enjoy all 21 of these design patterns at their leisure in the open-access paper. The key takeaway is that the authors are looking at cells not as poorly designed conglomerations of haphazard parts that some blind tinkerer cobbled together from whatever pieces of stuff were available, but as collections of elegant solutions to real problems familiar to engineers. It represents a noteworthy step toward design thinking in biology from an unexpected source.
Motivation for the Paper
In a video within the paper, Dr. Sauro from the Bioengineering Department explains what motivated the paper. He begins his answer by holding up a copy of Bruce Alberts’s textbook Molecular Biology of the Cell, a thick tome with 1,500 pages.
We started thinking: Is there any way we could abstract this information at a higher level, to help us comprehend what’s going on in a cell? And we were struck by this other book, which is totally different, Design Patterns. It’s a famous book in computer science by a so-called Gang of Four. It’s an interesting book because it describes how to solve complex problems in a sort of simplified way. And we thought: Is there was any way to marry this book with the Alberts book? That’s basically what motivated us to write this paper.
Following the order of the Design Patterns book, the authors divided systems in molecular biology into the same three basic categories: creational (such as the synthesis of a protein), structural (such as a phosphorylation cascade with inputs and outputs), and behavioral (such as a relaxation oscillator).
From this outline, the authors correlated the computer scientists’ design patterns with their actual implementations in cells. The implementations look like logic diagrams in circuit design. Mechanisms can be quite different, Sauro explains, and yet the underlying design pattern can be the same when examined at a higher level.
Importance of the Paper
Dr. Sauro feels the paper is important for a number of reasons. It provides a new way of communicating ideas in molecular biology, so that computational theorists and experimentalists can understand each other. Another benefit of the approach is to motivate other biochemists to build on their scaffolding of design patterns. This assumes many more engineering solutions can be identified; indeed, Sauro hopes others will help construct a searchable database of design patterns. Machine learning, then, could recognize patterns in newly identified networks in living organisms, expanding our understanding cellular networks. This would be very helpful for complex signaling networks, for instance, when it is hard to determine what is going on. Machine learning could compare known design patterns with the input/output behavior of the components, leading to an “Aha!” moment that untangles the complexity into a recognizable logic diagram.
Sauro credits primary author Steven Andrews for the clear and readable form in which the paper was presented. He hopes many scientists will read it, because it covers a wide range of biology and should interest all biologists — and, we would add, engineers. It is a springboard for ideas that also might interest those preparing for the next CELS conference.
Design patterns are recurrent solutions to commonly encountered problems. All biological cells encounter the same problems of how to construct the biochemical components that they are built from, how to connect those components together into useful reaction networks, and how to use those reaction networks to animate life.
The authors are quick to acknowledge certain predecessors in biological design thinking.
The idea of understanding cellular systems in terms of functional parts is of course not new. For example, Hartwell et al. argued for a modular view of cell biology, Del Vecchio et al. emphasized the central roles of control mechanisms, and Khammash’s group has focused on mechanisms that provide integral feedback control. In contrast to these and other works, our focus is larger, covering a wider swath of cell biology mechanisms. Also, our perspective is subtly different. Rather than focusing on a particular biological topic, our emphasis is on the development of a catalog of the solutions that cells have evolved to solve specific problems. This design pattern concept is useful for abstracting a broad range of cell functions into a manageable set of distinct patterns, enabling one to better see parallels and
Future of the Design Pattern Approach
Clearly, design thinking is a fruitful heuristic for discovery. But what about the “interlinked and hierarchical design patterns” mentioned next? Could those evolve? In the Illustra film Darwin’s Dilemma, such hierarchical patterns (exemplified in the body plans of the Cambrian fauna), are shown to resist Darwinian approaches because they require top-down design, as with a blueprint or logic diagram before assembly begins. Is this not the case with all “design patterns”?
The authors grant too much creativity to the neo-Darwinian mechanism. They assume that problems motivate their own solutions in biology:
Going even farther afield, one can speculate about life on other planets, where again the same problems would likely arise, and again would necessarily be addressed with many of the same solutions. This suggests that the design patterns listed here, along with others not addressed, could be reasonably considered universal principles of life.
Most likely this kind of speculation will wither on its own as the successors of Bruce Alberts add more pages to molecular biology textbooks. If, as the authors conclude, those involved in simulating cells will refer to a database of design patterns in their multiscale modeling, it should become increasingly clear that cells resemble engineered masterpieces. Darwinese would then decline as superfluous words in future research projects focused on design patterns.
