Fragmented language: the liminal space

The earliest complete sentence we can read in any alphabet was carved into a tiny ivory comb and said, “May this tusk root out the lice of the hair and the beard.” We can search for profundity here but the closest thing we’ll get to meaning is that people in around 1700 B.C.E. also suffered from unkempt hair and parasitic infections: the legibility is the miracle, the content is the mundanity. That’s not much, and it’s even less when you consider the thousands of years of life that came before anyone wrote anything down. And yet, there is an innate understanding between us and the person who was doing the grooming: written language forms the connective tissue between individuals to spur collective understanding or action. That’s extraordinary but incomplete.

The sentence is woefully underrepresentative of life at the time. Even if you took all the writing in the world and stuffed it into someone’s brain – like we did with language models – think of everything it wouldn’t know. It wouldn’t know small things like your breakfast or the last six thoughts you had or what the view from your living room window looks like. It would miss out on big things too: the shape of a family argument, the weight of a silence between friends. Our snapshots of the world, recorded through writing or photographs or other media, are extremely powerful, but they’re the visible tip of an iceberg of collective action, pressure, and responsibility.

An Ocean Between Shores

We like to look at biological components as individually acting or at least as stable states. Where economics likes to believe in the homo economicus, the rational agent, biology believes in the agency of the variant. Immunology looks at cell type or surface protein distribution to determine immunological response. Synthetic biology likes to look for a gene that can make one species artificially advantaged over another. Our biological understanding is essentially that there’s a normal – healthy, coded correctly, undisturbed – and then something changes it.

These phenotypes, these measurable individual states, are not permanent, they’re local optima. They’re legible recordings of a series of insults, perturbations, and departures that result in a sustained state. But the individual cells also maintain a quiet overhead, molecular flexibility, to adapt to a variety of reasonable circumstances, informed by their lineage and other genetic acquisitions. Evolution is explored in these margins: how efficient can something be and still be prepared for drought, cataclysm, even excess.

Rational agents can tell us a lot about rational agents. But they can’t tell us much about how they came to be shaped that way, or how they relate to each other. They operate, by definition, independently; things get spooky when you start to get people talking to each other.

Defects present flashing lights: we’ve identified artifacts that say someone with a p.Glu6Val on the HBB gene is going to see their hemoglobin force rigid red blood cells to stack into sickle shapes and cause pain and organ damage. But, again, that tells us about the defect-as-agent. We’re not so knowledgeable about how prolonged COVID response will cause T-cell exhaustion and a nebulous set of symptoms year after the infection.

The liminal space between individuals – words, currency, exchange – shapes collective action. In cellular terms, signaling cascades, peptides, and enzymes accumulated over time and space are the liminal space, promoting and maintaining these “silent functions” maintained in the genome. Our focus has been the individual: genomic, transcriptomic, and proteomic are tractable and neat in investigating the individual, and present real findings contextually mapped across cell states. They are tractable: we cannot overstate the value of tractability in science. But it’s also severely limited, especially in complex and systemic indications, the kinds of problems that we’re most interested in.

God is a Gerund

These complex problems are the most vexing because they operate on timescales that are both archaeological and immediate. Returning to the sickle cell anemia example, it happens that the specific mutation both causes a painful and serious illness while providing significant resistance to malaria. Creatinine is a signifier of kidney function. HDL and LDL ratios are standard measures of circulatory health. None of these form in a day: sickle cell is the result of thousands of years of evolution, kidneys lose function based on acute insult or progressively, and your diet plus individual propensity modulate your cholesterol.

Biology isn’t static. It’s a process and a response. And fundamentally, that process is a reflection of the conditions, environment, and competition faced by the biological entity. The external world is what we eat, what we know, how we get it, and sometimes what kills us. We may have been homo economicus once, but we’ve changed, man.

In molecular biology, “metabolomics” is often the discipline that’s tasked with filling that void. Strangely, that’s its blessing but also its curse: it represents a fathomless breadth of interaction while never really sticking. It’s a victim of its own size: it can measure Plasmodium repetitively eating cells and releasing toxic byproducts, lipids modulating membrane plasticity, the accumulation of PFAS in newborns, and someone’s acetaminophen dose, all in one sample, each with a substantial implication on the biological trajectory of the individual.

Our understanding of biological-environmental interaction is caught in a paradox. Everyone – a salesperson, an executive, a scientist – wants a smoking gun. Metabolomics is a smoking gun. Often only a smoking gun. This is what makes metabolomics structurally different from its sibling disciplines. Genomics, transcriptomics, and proteomics map cleanly onto function; their targets are finite alphabets, their methods standardize, their outputs reduce to legible categories. Metabolomics doesn’t. The chemistry is too diverse, the signals too context-dependent, the same molecule doing too many things in too many places. This isn’t a failure of methodology waiting for better tools. It’s a feature of what small-molecule chemistry is.

The Search for Esperanto

Metabolomics is necessarily orthogonal to our biological terms and that’s a good thing. Where a road map is good, a traffic map is better, and metabolic flux demonstrates the independent choices that systems are making with respect to their environments and other agents. With that comes a cost: the abstraction layer isn’t there yet and the interpretive analysis exists in a separate place, nearer the lab and the instrument.

In 1887, L.L. Zamenhof constructed the language of Esperanto to “enable the learner to make direct use of his knowledge with persons of any nationality, whether the language be universally accepted or not; in other words, the language is to be directly a means of international communication”. Getting agents to speak, collaborate, and change their behavior is an existential problem, tangible in human interactions, but essential to how all biological systems form. Why, then, is this language merely a novelty, a Jeopardy question, and not widely spoken? Because it’s irredeemably reductive.

Language forms not just the basis of our interactions but a template of the contexts of our lives. It is both the mundanity of a comb for lice and the seriousness of “fire” in a crowded theater. It is multipurpose and culturally reflective. It expresses imperatives, questions, and descriptions in order to get distinct biological entities to relate more closely with each other. It is an enormous liminal space between things.

Small molecules and their elements are the natural analog to our smaller-scale forms. This resists abstraction, as language resisted abstraction into Esperanto. While this might seem like a niche scientific problem, it has real world implications: until we understand the language, the coordination, and the contexts of our biological systems, we won’t know them at all. We can describe things with words but we can’t describe words with things.