Can Proteins Record Earth’s Oxidation History? An Evolutionary Clue from Open Sequences and Thermodynamics

When reading the early history of Earth, people typically think of rocks, minerals, and sediments. Another line of inquiry starts from the biological materials that still exist today: could the chemical composition of proteins also preserve clues of the joint changes of environment and evolution? This is not treating modern proteins as pristine, preserved ancient specimens, but rather using sequences, phylogenetics, and chemical calculations to establish inferences that can be cross-checked by other evidence.

Researchers including Jeffrey Dick proposed the study titled "Evolutionary oxidation of proteins in Earth’s history," and released the corresponding genoGOE program and data, as well as versioned data archiving. From the public materials, it is evident that the analysis links modern genomes, various age groups of genes, and multiple ancestral protein sequences, while providing methods for generating plots and source data. It gives readers an opportunity to ask: how exactly is a chemical trend spanning vast geological time constructed from sequences available today?

The appeal of this research direction lies in placing Earth's chemistry and molecular evolution into a comparable framework. Its limitations are equally clear: the reconstructed sequences are not directly observed ancient samples, gene ages are not clocks without error, and chemical quantities cannot be directly converted into the atmospheric oxygen content of a specific era. If these intermediate steps are ignored, new proxy clues can easily be written as historical conclusions that exceed the evidence.

The public abstract of the research primarily describes the observation as: the average oxidation state of carbon in various data sets shows an upward trend across a timescale of billions of years. The authors also preserve important nuances, noting that while larger changes and the Great Oxidation Event can only be said to roughly overlap, they cannot be said to be precisely synchronized; a later rise shows different patterns in the analysis of gene ages for unicellular and multicellular eukaryotes. These limitations make the research closer to a testable hypothesis rather than a simple history shared by all life.

The thermodynamic explanation proposed by the authors links chemical trends to conditions such as Earth's gradual oxidation and long-term cooling. Compatibility with an explanation does not equal the exclusion of all other causes; especially since phylogenetics, spatial correlations, and sample origins still limit comparisons. For the reader, the key is not just seeing the direction, but retaining the exceptions and uncertainties the researchers themselves identified. If these are removed, leaving only the narrative that Earth's oxygen rewrote all proteins, it turns a study requiring cross-evidence verification into a conclusion it has not proven.

The Carbon Oxidation State of Proteins is a Compositional Quantity

Proteins are composed of amino acids, and different amino acids have different elemental proportions. After converting sequences into chemical compositions, quantities such as the average oxidation state of carbon can be calculated. Dick's early published full-text research has discussed the connection between this type of compositional description and issues regarding proteins, the environment, and evolution. The oxidation state here does not equal the degree to which proteins were damaged by oxygen, nor is it an antioxidant indicator in a health context.

For the general reader, the most important thing is to distinguish the object of measurement. The study compares the chemical composition determined by sequences, not the oxygen concentration directly measured in a cell at a given moment. Different proteins may have different compositions due to function, structure, cellular location, and evolutionary history. Therefore, after observing different chemical quantities, it is still necessary to explain which differences might be related to the environment and which stem from the limitations of the organisms themselves.

Averages can also hide internal variations. The protein composition of an entire genome and the ancestral sequence of a specific class of enzymes are not the same unit of analysis. The former may be influenced by the proportions of different protein classes, while the latter may be more susceptible to influences from functional and sequence reconstruction assumptions. If different materials show similar trends, it can increase the importance of the research question, but it still requires checking whether they share the same sources of bias.

How are Modern Sequences Placed into Time?

The data listed in the public repository include modern microbial genomes, age classifications of eukaryotes, and ancestral sequence materials such as nitrogenase, thioredoxin, and Rubisco. Different materials provide different temporal clues and cannot be treated as the same set of direct measurements. Researchers need to link chemical compositions with relative or estimated time based on existing phylogenetics, sequence alignment, and age estimation.

Ancestral sequence reconstruction can be understood as estimating the amino acids more likely to appear at a certain ancestral position under known descendant sequences and evolutionary models. It is not reading a complete protein directly from rock, nor does it mean every position is equally certain. If different models or trees produce different sequences, these differences should be checked for how they propagate to the final chemical quantities, rather than just displaying the most convenient ancestral version for calculation.

Gene age classifications also have their own limitations. How far back a certain gene can be traced in a database is influenced by sampling, sequence identification, and classification methods; it does not necessarily equal the moment it was actually born. If distantly related sequences are difficult to identify, older genes might be grouped into younger clusters. These issues do not make the methods lose their utility, but they require researchers to treat age inferences as data with uncertainties, rather than fixed historical coordinates.

Thermodynamics can Provide Constraints, but cannot Monopolize the Evolutionary Story

Tools used in the public analysis include protein composition calculations and geochemical thermodynamics software. Thermodynamic methods can help compare material requirements and reaction relationships under different chemical conditions, but real biological systems still involve regulation, function, dynamics, and historical contingency. Jumping directly from the feasibility of energy or matter to the inevitable evolution of a certain sequence still lacks the intermediate processes that need verification.

Therefore, a better way to read the work is to ask what the model excluded and what it retained. The temperature, chemical conditions, and reference species in the model are choices made during analysis; whether they are close to the actual subjects of study requires separate discussion. If the conclusion remains stable under reasonable alternative settings, its credibility can increase; if it is highly sensitive to a specific parameter, that dependency should become part of the research results rather than being treated as a technical detail not worth explaining to the reader.

Environment and life can also influence each other. Biological metabolism changes the environment, and the environment may in turn affect the resources available to organisms and the conditions of selection. Writing changes that roughly overlap in time as a one-way causality would lose this reciprocal relationship. Earth's oxidation, cooling, and protein composition can be placed in the same research question, but one cannot judge which factor explains all the changes solely based on similar trends.

Reproducible Programs are the Entry Point to Opening the Question

genoGOE's public documentation lists the program environment, required packages, and how to generate primary and supplementary charts; each chart can also output its source data. This allows other researchers to check how data processing and graphics are formed step-by-step. Versioned archives help fix the content used in a specific analysis, preventing subsequent updates from causing comparisons to lose a common baseline.

However, "program publication" and "results independently reproduced" must be separated. Having read the instructions does not equal successfully installing the environment; running the program does not equal verifying all original sequences; reproducing the graphics does not equal validating the interpretation of the ancient environment. True re-analysis requires recording versions, inputs, methods, and points of deviation from the original process, and checking which differences would change the research judgment.

Having many data sources does not necessarily mean the evidence is fully independent. Different analyses might use the same sequence databases, age frameworks, or chemical conversion methods; if a common assumption is biased, multiple seemingly consistent lines may be affected together. The value of the public materials lies in providing an opportunity to identify these common parts and allowing subsequent research to design truly different checks, rather than just redrawing similar charts.

Dialogue with Rock Records is the Value of Proxy Indicators

Evidence from rock geochemistry and protein composition has different preservation, sampling, and inference limitations. Putting the two together is not about declaring that a new molecular method replaces old geological methods, but about using the uncertainties in each to test the other. If estimated ages do not align perfectly, it may be necessary to check time resolution, samples, and models; if trends are similar, it should also be further asked if there are common causes, rather than immediately declaring that the same answer has been found.

This research relationship can form a two-way question: Can geological records limit the environmental conditions in sequence models? Can sequence analysis point out geological periods or biological groups worth re-examining? Both sides need to maintain the ability to present counter-examples. The most meaningful moment for a new method is not necessarily when it perfectly aligns with the existing story, but when it allows a problem previously seen as simple to reveal new dimensions.

For science communication readers, it is also necessary to note that "oxidation during evolution" has no direct equivalence to personal diet, disease, or health claims. Analysis of sequence composition across geological time cannot be used to judge whether a certain protein is healthier today, nor can it support the efficacy of supplements. Similar wording does not mean the research questions are the same; mixing them up will cause a piece of fundamental research to lose its original scale and meaning.

Whether proteins can become another clue for reading Earth's history still depends on the joint verification of verifiable sequences, models, and geological evidence. What is worth looking forward to is researchers putting forth their methods and materials, allowing historical inferences to be rerun, refuted, and corrected, rather than allowing all changes in life to be condensed into a single, aesthetically pleasing, but unquestionable chemical curve.

Data Sources

• Dick et al.: Evolutionary oxidation of proteins in Earth’s history, research abstract and entry to accepted version
• genoGOE: Original research program, data sources, and reproduction instructions
• genoGOE v1.1.0: Versioned program and data archive
• Dick: Average oxidation state of carbon in proteins, published full-text research
• Dick et al.: Using thermodynamics to obtain geochemical information from genomes, original research abstract