← Latest papers
🧬 biology

Provenance-aware observations of mitochondrial organization

This paper introduces "armosis," a quantitative framework measuring the progressive reduction in theoretically possible biological entities across organizational levels, and demonstrates its application using a provenance-aware knowledge graph of human mitochondrial genomes to reveal increasing constraints from codons to the complete electron transport chain.

Original authors: David A Stumpf

Published 2026-07-29
📖 7 min read🧠 Deep dive

Original authors: David A Stumpf

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Great Biological Filter

Imagine you are walking through a massive library where every book represents a possible way a living thing could be built. In this library, the shelves stretch infinitely in every direction, filled with every conceivable combination of letters, words, and chapters. This is the world of "theoretically possible" biology. If you took every single letter of DNA and shuffled them randomly, you could create trillions of different creatures. But here's the twist: if you actually walk out into the real world and look at living things, you won't find them scattered randomly across those infinite shelves. Instead, you'll find them clustered tightly in just a few tiny corners.

This paper explores a fascinating mystery: Why does nature seem to ignore almost all the possibilities? Scientists call this phenomenon armosis. Think of it as a giant, invisible filter that biology passes through at every step of construction. It starts with the tiny building blocks (DNA letters), moves up to the instructions (genes), then to the machines those instructions build (proteins), and finally to the big factories (cellular complexes). At each step, the filter gets stricter, throwing away almost all the random combinations and keeping only a very specific, limited set of working designs. The paper asks: How much of the "possible" universe do we actually see in real life, and does this filtering happen in a predictable way?

The Story of the Shuffled Deck

In this study, a researcher named David Stumpf decided to investigate this filtering process by looking at the mitochondria—the tiny power plants inside our cells. He didn't just look at one person; he gathered a massive collection of 53,166 complete mitochondrial genomes from the GenBank database, which is like a giant public library of genetic codes. To make sense of this mountain of data, he built a special digital map called a knowledge graph.

Imagine a knowledge graph as a giant, interactive web of sticky notes. Each note is a piece of data (like a specific DNA sequence), and the strings connecting them show exactly how they are related. The special thing about this map is that it keeps "provenance," which is just a fancy word for keeping a perfect paper trail. It remembers exactly where every piece of information came from and how it was assembled. This allowed the researcher to trace a path from a single DNA letter all the way up to a fully functioning protein machine, without losing track of the journey.

The Great Compression

When Stumpf started tracing these paths, he found something incredible. He discovered that while there are billions of theoretically possible ways to arrange the DNA letters for a specific protein, nature only uses a tiny fraction of them. He measured this using a new concept called armosis, which is basically a score that tells you how much the "possible" universe shrinks down to the "observed" reality.

The results showed that this shrinking happens in stages, like a funnel getting narrower and narrower:

  • Step 1: When looking at how DNA letters (codons) turn into protein instructions, the filter is already working. The armosis score was 1.5, meaning the number of observed options was already much smaller than the total possibilities.
  • Step 2: As he moved up to how these protein instructions assemble into larger machines (like Complex I, a part of the cell's energy system), the filter got much stricter. The armosis score jumped to 10.7.
  • Step 3: By the time he looked at the entire electron transport chain (the whole energy-making system), the score was a massive 30.9.

This means that as you build bigger and more complex biological machines, the number of working designs nature actually uses becomes a microscopic fraction of what could exist. It's as if you had a deck of cards with infinite suits, but every time you tried to build a house of cards, you were only allowed to use a handful of specific cards.

The "Checkerboard" of Life

One of the most surprising findings was that these limited designs aren't random. The researcher found that certain protein parts tend to "stick together" like puzzle pieces that only fit with specific partners. Even though the genes for these proteins might be far apart on the DNA strand, they travel in pairs or groups.

He described this pattern as a "reticulate checkerboard." Imagine a checkerboard where the black squares are the most common combinations of genes, and the white squares are rare ones. Most people have a "black square" pattern, but there are a few "white square" variations. Crucially, the study showed that these patterns are not just local; they are global. For example, the genes for ATP6 and ATP8 (which overlap on the DNA) do stick together, but they are actually ranked only 33rd in terms of how strongly they associate with each other. Many other genes that are far apart on the DNA have even stronger "friendships" and travel together more often.

The study also looked at whether we would keep finding new, wild combinations if we sequenced more people. Using a method called saturation analysis, they found that we are approaching a limit. They estimated that the total number of distinct protein-gene combinations in humans is finite, around 12.5 million (based on their model). They calculated that if we sequenced about 81,842 genomes, we would have found 95% of all the common combinations, and by 125,811 genomes, we would have found 99%. This suggests that nature has a "finite repertoire" of designs. We aren't discovering endless new types of mitochondria; we are just finding the rare, long-tail variations of a limited set of templates.

What This Means (and What It Doesn't)

The paper suggests that biology is not a chaotic free-for-all where anything can happen. Instead, it is a highly constrained system where functional requirements act as a strict editor, cutting out almost all the "what-ifs" and leaving only the "what-works."

However, the author is careful to say that this study does not explain why this happens. It doesn't tell us the specific biological mechanisms (like specific chemical reactions or evolutionary pressures) that force nature to choose these specific patterns. It simply measures that the patterns exist and quantifies how strong the constraints are. It's like noticing that a city only has traffic lights at certain intersections and never at others; the study maps the traffic lights perfectly but doesn't explain the city planning committee's meeting minutes that decided where to put them.

The study also rules out the idea that these patterns are random or that they keep expanding forever. The data shows a clear, finite limit to the variety of mitochondrial designs we see in humans. While the study focuses on human mitochondria, it hints that this "constrained assembly" might be a general rule for how complex biological systems are built, suggesting that life explores a vast space of possibilities but only ever inhabits a very small, highly organized neighborhood.

In short, this paper gives us a new way to look at life: not as an endless explosion of variety, but as a carefully curated collection of working models, where every step up the ladder of complexity filters out more and more of the impossible, leaving us with the elegant, limited designs we see today.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →