In the quest to make the theory of evolution mathematically rigorous, scientists develop ever-more-complex models to describe how life’s machinery might change over time. The 2010 PNAS paper, “Mutation-selection models of coding sequence evolution with site-heterogeneous amino acid fitness profiles” by Nicolas Rodrigue and his colleagues, is a prime example of this effort. The authors present a sophisticated statistical model to better understand the interplay between mutation and natural selection at the level of protein-coding genes. Proponents of evolutionary theory might point to such work as a sign of a mature, advancing science. However, a critical analysis reveals that the paper, while a feat of modeling, fails to provide any support for the grand claim of unguided, molecules-to-man evolution. Instead, its findings inadvertently highlight the impotence of the mutation-selection mechanism to create new information and, when properly interpreted, provide powerful evidence for intelligent design and the ongoing reality of genetic decay.
A Fair Summary of the Research
The stated goal of Rodrigue et al. is to create a more realistic model of how protein-coding DNA sequences evolve. Previous models were often too simplistic, either assuming that every position in a gene is under the same type of selective pressure (a homogeneous model) or that every single position is unique in a way that is too complex to manage (an overparameterized model).
The authors’ solution, which they call the MG-MutSelDP model, is to recognize a biological reality: for a protein to maintain its function, different positions (sites) along its amino acid chain have different levels of tolerance for change. Some sites are highly constrained, perhaps allowing only one or two specific amino acids, while others are more flexible. Their model uses an advanced statistical method (a Dirichlet process) to infer these site-specific “amino acid fitness profiles” directly from alignments of existing, related genes.
The key, narrow findings of the paper are twofold. First, their model successfully distinguishes between the underlying tendencies of mutation (e.g., a bias toward certain nucleotide changes) and the effects of natural selection, which filters those mutations. Second, the model confirms that selective pressures are indeed highly variable across the sites of a protein and that these pressures are overwhelmingly dominated by purifying selection—the process of eliminating harmful mutations to conserve a protein’s existing function.
Modeling Maintenance, Not Creation
While the authors’ model is an elegant piece of statistical work, it is critically important to understand what it does and does not do. The entire analysis is founded on a central, unstated assumption that makes it irrelevant to the question of evolutionary origins.
The “Assume a Gene” Fallacy: The study begins with data sets of pre-existing, functional, and profoundly information-rich genes like globins and rhodopsins. The model’s purpose is to analyze the patterns of variation within these already existing gene families. It is an exercise in reverse-engineering the pre-loaded functional constraints required to keep a globin protein working. The model offers absolutely no explanation for the origin of the globin gene itself, nor for the origin of the complex, specified information that defines its function and creates the very “fitness profiles” the authors seek to measure. It explains the fine-tuning of existing technology, not the origin of the technology. This is like studying the wear-and-tear patterns on a jet engine to understand its maintenance requirements, and then claiming you’ve explained how the engine designed itself from raw metal.
A Portrait of Genetic Entropy in Action: The most powerful result of the paper is a stunning, albeit unintentional, confirmation of the principle of genetic entropy. In Figure 3, the authors present the distribution of fitness effects for all possible mutations. Their model calculates that for the globin dataset, 87% of potential mutations are either deleterious (harmful) or neutral. For the rhodopsin dataset, this figure rises to a staggering 92.9%. This is a direct mathematical portrait of a system being relentlessly bombarded by destructive errors.
This finding demolishes the popular notion of mutation as a creative engine for evolution. Mutations are overwhelmingly a corrupting force. The model shows that natural selection’s primary role is not to innovate, but to act as a quality-control filter, purifying the gene by removing the constant influx of harmful changes. This is not a mechanism for ascent; it is a damage-control system for a process of decay. The authors have not modeled evolution; they have modeled devolution and the conservation of a decaying system.
Quantifying the Barrier to Innovation: The model’s inference of “site-specific fitness profiles” is simply a technical term for what design proponents call “functional specificity.” By showing that only a small subset of amino acids is permissible at most sites, the authors quantify the immense challenge of creating a functional protein by chance. The search space for a new functional protein is not a wide-open prairie of possibilities but a treacherous labyrinth with a single, narrow path to the goal. A blind, unguided search is statistically doomed to fail. The authors’ model mathematically confirms the extreme rarity of functional protein sequences, a cornerstone of the intelligent design argument articulated by biochemist Douglas Axe.
An Alternative Explanation: A Designed System Undergoing Decay
The evidence presented by Rodrigue et al. does not require the narrative of unguided evolution. In fact, it fits far more naturally within a framework of intelligent design and a recent creation.
Inference to the Best Explanation: The paper’s core premise is that proteins are governed by “site-heterogeneous amino acid fitness profiles”—in other words, specified functional constraints. According to the principle of vera causa, we should seek a cause that is known from our uniform and repeated experience to produce the effect in question. What is the only known cause of specified functional constraints? Intelligence. An engineer specifies precise materials and tolerances at every point in a machine for it to function correctly. The “fitness profiles” the authors discovered are not the product of a blind process; they are the engineer’s specifications, embedded in the digital code of the genome.
A Model of a Created “Kind”: The model is applied to alignments of similar genes (e.g., globins, rhodopsins). From a genealogical perspective consistent with the biblical account, these gene families do not represent a universal tree of life stretching back billions of years. Instead, they represent the designed diversity and subsequent modification within an original, created “kind.” The underlying “fitness profile” is a reflection of the robust, optimal design template created by the Engineer. The mutations accumulating within the population are the corruptions introduced into that perfect system after the Fall. The model, therefore, is not tracking the deep-time evolution of a gene, but the recent, post-creation diversification and decay of a single, masterfully designed biological system. This framework also accounts for the data without needing to invent rescuing devices to explain away the empirically measured fast molecular clocks, which consistently point to a history of only a few thousand years.
Conclusion
In their sophisticated attempt to build a more realistic model of gene evolution, Rodrigue and his colleagues have inadvertently forged a powerful tool that demonstrates its core weakness. By presupposing the existence of information-rich genes and then modeling the real-world forces acting upon them, they have shown that the mutation-selection mechanism is a process of overwhelming conservation and decay, not creative innovation. Their model mathematically confirms that mutations are predominantly harmful and that natural selection’s primary role is to weed out this damage, preserving existing function for as long as possible.
The “site-specific fitness profiles” they meticulously model are not evidence of an unguided process but are better understood as the functional specifications of a master programmer. When viewed through the lens of rigorous historical science, which demands an appeal to a causally adequate source, the data points directly to intelligent design. The paper, far from supporting the grand narrative of molecules-to-man evolution, provides a detailed, quantitative picture of created, information-rich systems fighting a losing battle against the universal law of entropy.
Leave a Reply