What Molecular Mimicry Is, and Why It Matters for Autoimmune Disease
When a pathogen's proteins resemble our own closely enough, the immune system can learn to attack both. This is the mechanism at the center of our thesis.
A case of mistaken identity
The immune system recognizes threats by molecular shape. Most of the time this works well — but some pathogens carry proteins structurally similar enough to our own that the immune system, once trained to attack the pathogen, goes on attacking the look-alike in our own tissue long after the infection has cleared. This is molecular mimicry, and it is one of the leading mechanistic explanations for how autoimmune disease can begin.
Why it's hard to see
The difficulty is scale. The human proteome and a pathogen's proteome each contain thousands of proteins. Finding which pairs are similar enough to plausibly cross-react cannot be done by inspection — it requires comparing every protein in one system against every protein in the other, at a resolution fine enough to catch partial matches, not only identical sequences.
Our approach, in preliminary form
Our own work in this area began with exactly that comparison: a proteome-wide alignment between a pathogen — Acanthamoeba, an organism with documented links to chronic and autoimmune-adjacent conditions — and the human monocyte-macrophage system, the immune cells most directly involved in first response and antigen presentation. The results are preliminary, and we treat them as such: a starting map of where cross-reactivity is structurally plausible, not a finished causal account. A proteome-wide alignment can show where the immune system might reasonably confuse pathogen for host; it cannot by itself show that this confusion causes disease in a given patient. That link has to be established separately.
Where this leads
This preliminary mapping underlies ANA, our system for categorizing molecular-mimicry patterns at proteome scale, built to make this kind of comparison systematic rather than one-off. To our knowledge, no equivalent tool operates at the same scale today. We will publish further findings as the categorization work moves past this preliminary stage.