In December 2025, Molecular Systems Biology marked its 20th anniversary with a special editorial that reflects on the field’s development since 2005 (Bheda et al. 2025). Systems biology is an approach to studying living systems that assumes hierarchical, top-down design. The piece, authored by the journal’s editors and several contributors, shares personal perspectives on where the field stands today — and where it is headed. Ruedi Aebersold, the first contributor, states, “the first 20 years of MSB were grand; the next 20 years will be grander.”
I too am optimistic about the field’s future. My optimism comes specifically from how powerfully top-down design has succeeded in giving us the complex systems of the modern world. Top-down design prunes the vast search space of possibilities through an understanding of overall system function, performance requirements, and constraints. We’ve all benefited enormously from the top-down design that has led to data networks, computers, smartphones, and countless other technologies.
A “Reverse” Application
I believe biology needs a comparable “reverse” application of these principles. To learn more about the approach I suggest, check out a paper I co-authored, “A Model-Based Reverse System Engineering Methodology for Analyzing Complex Biological Systems With a Case Study in Glycolysis,” reviewed here. This approach I think can help us decode and make sense of biological systems in ways that purely bottom-up approaches struggle to achieve.
As an advocate of intelligent design, I see systems biology — with its inherently top-down philosophy — as a natural and seamless fit with ID. In contrast, attempts to reconcile systems biology’s top-down reality with the bottom-up nature of Darwinian evolution always feel a bit forced. For example, I’ve not seen serious investigation of the waiting times required to achieve the top-down design of even a simple system. There is no consideration of how many single-step mutations would be necessary, nor any estimates of the amount of coordination required. I think this is because the problem is actually too hard. If one protein cannot evolve (Dilley et al. 2023; Axe 2004) from scratch, what’s the probability of ten evolving at the same time in a coordinated manner towards a goal that a blind process cannot see?
In the rest of this post, I’ll offer some commentary on this editorial — which does an excellent job of surveying the field — and focus specifically on how intelligent design can contribute several unique and valuable insights to systems biology.
The Three Phases of Systems Biology
Aebersold divides the field’s development into three distinct phases:
Phase 1: High-Throughput Molecular Biology
The birth of the field was driven by the explosion of “omics” technologies — genomics, transcriptomics, proteomics, metabolomics. These tools generated vast datasets that quickly revealed the limits of the classic “one gene–one protein–one function” paradigm. These data showed that biology does not resemble a Rube Goldberg assemblage of suboptimal, repurposed parts, but rather a highly optimized system in which each component appears finely tuned for its particular place in the whole (exceptions probably exist, but they are exceptions, not the rule).
What intelligent design offered then, and still offers today, is a framework that actively encourages assumptions of optimality, function, and purposeful design when interpreting big data. By presuming function and optimization as default expectations, one is justified in investigating elements that are otherwise (from a Darwinian perspective) often presumed to be unimportant. Some such elements include non-essential genes, so-called “junk” DNA, low-abundance mRNA, long non-coding RNAs, and alternatively spliced protein isoforms.
I believe that over the next twenty years, some of the big successes in systems biology will come from assuming good design and optimality around key top-down design requirements when looking at big datasets.
Phase 2: Network Biology
After the flood of big data came in, the field shifted toward trying to map all the data, which led to the discovery of biological networks. For example, by integrating transcriptomic data with genomic data, researchers mapped interactions between transcription factors and their target genes, giving rise to transcription regulatory networks.
However, there’s a big difference between seeing static interactions (transcription factor 1 targets gene 1 and gene 2), and being able to make sense of what’s happening and why the network has a given structure. A great example of the higher level of understanding is the presentation of transcription network motifs by Uri Alon, one of my personal scientific heroes. In his book Introduction to Systems Biology (Alon 2019), Alon describes how to find patterns of connections, which helps us understand how and why the network is structured in a particular way. For example, after comparing the E. coli transcription network with a random network, he uncovered motifs such as negative autoregulation (which accelerates response times and reduces noise), the feed-forward loop (both coherent type-1, acting as a sign-sensitive delay element, and incoherent type-1, enabling response acceleration and pulse generation), and single-input modules (coordinating groups of genes). Each of these has particular functional advantages which explain their particular use at a point in the network.
Here again, I think that intelligent design has something unique to offer the field. Several preliminary analyses, including the one mentioned above, have discovered that the network motifs or design patterns in biology are often the same ones used in human engineering. The reason is that when faced with the same physical constraints and design objectives, designers converge on similar design patterns. This is why most cars have four wheels instead of five or three, even though they are made by different designers. Knowing this can be helpful for reverse engineering (aka doing biology). We can look for known engineering-design motifs in biology. When we find them, we often already know from the experience of human engineering why they are necessary, and this can help us understand the architecture of the biological system.
Phase 3: Complex Adaptive Systems (CAS)
I found Aebersold’s notes on Phase 3 particularly interesting, as I’ve been thinking about several of its constituents (adaptability, distributed control, robustness), but I had not been thinking about these together as the next phase. What stands out is that Phase 3 is moving from static to dynamic descriptions of systems (hence the use of “adaptive”) where prediction starts to become more of a reality.
In this editorial another contributor, Joel Bader, writes that we know the parts, but what about the wiring? I think this is the sticky challenge for Phase 3. We have extensive knowledge of genes and proteins, but understanding their relationships as an integrated system remains elusive. My favorite article on this topic is Yuri Lazebnik’s classic, “Can a Biologist Fix a Radio?” (Lazebnik 2002). Lazebnik humorously yet powerfully illustrates the difficulty of deciphering complex biological systems using traditional reductionist approaches.
I believe part of the solution lies in better modeling of systems and greater integration of requirements analysis into biology. You need to have a solid qualitative model before you attempt to build a quantitative one. Accordingly, I want to see systems modeling become more mainstream in biology. Some progress is being made — for example, resources like Gene Ontology and AmiGO provide valuable structured knowledge — but they still fall short of true systems models. I now use OpCloud to build models of any biological system I’m investigating. It helps me organize data from research papers, construct a mental framework grounded in real systems principles and relationships, and easily abstract from low-level details to higher-level views while linking back to system requirements. But I’m one person; we need this on a much larger scale.
Biology and Optimal Design
Reading this editorial felt deeply affirming. The reflections on field development — especially the phases collection of big data → network mapping → adaptive dynamic systems — mirror thoughts I’ve had since the beginning of my work in this field around 2020. What unites all these phases is that the systems biology approach assumes what reductionist biology did not — and I would argue could not: complexity, usage of mathematical principles, usage of engineering principles, perhaps all of which could be summed up as making the assumption that biology exhibits optimal design.
References
- Alon, Uri. 2019. “An Introduction to Systems Biology.” Preprint. https://doi.org/10.1201/9780429283321.
- Axe, Douglas D. 2004. “Estimating the Prevalence of Protein Sequences Adopting Functional Enzyme Folds.” Journal of Molecular Biology 341 (5): 1295–1315.
- Bheda, Poonam, Jingyi Hou, Ruedi Aebersold, et al. 2025. “Molecular Systems Biology at 20: Reflecting on the Past, Envisioning the Future.” Molecular Systems Biology 21 (12): 1667–1673.
- Dilley, Stephen, Casey Luskin, Brian Miller, and Emily Reeves. 2023. “On the Relationship between Design and Evolution.” Religions 14 (7): 850.
- Lazebnik, Yuri. 2002. “Can a Biologist Fix a radio? — Or, What I Learned While Studying Apoptosis.” Cancer Cell2 (3): 179–182.