Recently, a team at Stanford University fished something new out of the vast, uncharted, and almost entirely unclassified world of genetic material sometimes known as biological dark matter. They called it an “obelisk,” a shell-less RNA of maybe 1,000 base pairs (shorter than any viral genome), which seems to self-organize into a rodlike shape. It appears to be the structural equivalent of a plant viroid or a fungal ambivirus, two other bits of self-replicating genetic material whose discovery widened the boundaries of what we know about microbiology. But the obelisks weren’t found in either plants or fungi. They were discovered in human intestines.

An obelisk, in other words, is a probably replicating entity, contained in a mere thousand or so letters of the genetic alphabet, with zero genetic similarity to anything known to exist already. What does it do? How did it get there? Does it cause disease? We have no idea. We first picked up the phone this year (the preprint announcing its discovery was published in January 2024 ¹ ). It is a complete stranger calling from inside the house. 

The thing about microbiology is that it’s really, really easy to miss. Through history, its least ignorable effects — fermentation, mold, rot, agues, poxes, and other pestilences— were attributed to various disparate causes: bad vapors, natural physical processes, moral failures, divine intervention. Demons were frequent causes of infectious disease, as were sinister gods. But the work of microbes was also credited to divinity. Early Western philosophers and alchemists had varying ideas on what fermentation was, but it was often seen as something as much spiritual as chemical. Jesus said, “The Kingdom of Heaven is like the yeast a woman used in making bread. Even though she put only a little yeast in three measures of flour, it permeated every part of the dough.” ²

You can ride these folk microbiologies pretty far. You can keep those who have died from disease far away from you without understanding exactly why that helps. And in your kitchen you can keep mothers of vinegar or slurries of foam for beer or bread on from batch to batch without knowing that you’re harboring great empires of yeast and bacteria.

As far as we know, the first person to see microscopic bacteria was biologist and microscope maker Antonie van Leeuwenhoek in 1676. Peering through his finest lens yet, he was shocked to realize that there were moving creatures down there. (He called them “animalcules.” ³ ) Nobody followed up on that for a couple hundred years. Ferdinand Cohn used more cutting-edge microscopes to study microscopic algae and to class bacteria by shape around 1870, which is about the time microbes rose as an object of scholarly interest. Robert Koch showed that a particular bacterium caused tuberculosis in rabbits, and that the same bacterium could then be found again in the sick rabbits: the mechanism underlying infection. And Louis Pasteur demonstrated that rot and fermentation don’t occur spontaneously, but require a living microbial source. So began the race to identify the organisms behind other diseases.

To help with research, Pasteur’s colleague, Charles Chamberland, developed a ceramic filter that they thought would sterilize water — the pores were too small for bacteria to pass through. It worked for many purposes. But within a decade, Martinus Beijerinck identified an infectious disease in plants, where the contaminating material seemed to be able to pass through Chamberland’s filter. They wondered: Could there be some pathogen that was smaller than a bacteria? 

What Beijerinck did not know was that he had found the most common kind of life on Earth. But all that was still a long way off.

First things first: Pick a bacteria and study it. Wait. How do you “pick a bacteria”? The status quo since the time of Koch and Pasteur is to isolate and culture: You take a sample and let it grow in a nutrient-rich medium, then pick out one colony, and put it in new media. Repeat and repeat. Even if your starting sample had 10,000 different germs in it, by growing just a little of it repeatedly, you will eventually have just one kind of germ, and the ability to make more of it.

This simple engine, plus improving microscopy, powered microbial characterization from the 1870s to the 1970s, ⁴ and led to astonishing progress considering how laborious it was to get anything done. Like animal taxonomy, microbes were classified based on observable traits. Unlike animal taxonomy, it was extremely difficult to observe any traits at all.

Under the microscope, you could observe the size of cells, their shape, their behaviors, what they grew on and how fast, what killed them, the chemical composition of their skins. If you determined a lot of these facts, you could approach something that almost resembled species identification, and that’s what diagnosis ran on for a long time — vast quantities of carefully maintained microbial gardens. 

There are problems in taxonomy based on simple observation. For instance, Escherichia coli, most famous of free-living bacteria, was assumed for a long time to be one well-understood narrow clade — even something akin to a species, as the Genus species name would have you believe. It had a known size, shape, metabolism, and behavior, and happened to be very common. Only within the last couple decades has genetic sequencing revealed that in fact some “subspecies” of E. coli have only 20% of their genes in common. ⁵ These very disparate microbes were hiding as one species under an umbrella of similar observable properties.

Back in 1898, Beijerinck had theorized that there existed an infectious agent even smaller than bacteria. He wasn’t able to culture this new type of pathogen, but he did give it a name: virus. At least with pathogens, you know one circumstance in which they will grow: inside a suitable host. Bacteriophages, viruses which attack bacterial cells, are easy. That’s how Félix d’Hérelle first cultured them in 1917: Grow the bacteria first, add the bacteriophage, and then look for clear spots where the phage has laid waste. For some time, that was the way to detect phages: by looking for absence. It took until 1938 and the advent of electron microscopes before anyone saw a virus. ⁶

Beijerinck thought that viruses were liquid: a contagium vivum fluidum, or a contagious living fluid. In fact, they are particulate parasites: a short genome in a protein shell that borrows everything else needed for reproduction from its hijacked host cell. Some of them are huge, the size of bacteria, so large that — for a virologist — they’re easy to miss, because a filter small enough to catch all bacteria catches them too. Most viruses are smaller, and they’re in all sorts of shapes: soccer balls, lemons, spaceships complete with landing pads, cooked spaghetti, uncooked spaghetti, wine bottles, bullets, teardrops. When the last common ancestor of all other life was new, viruses were old. ⁷ Across the world, viruses of eukaryotes, archaea, and bacteria outnumber their host cells by at least 10 to one. ⁸ They may be the most common reason cells die in nature.

Viruses are everywhere. But viruses that attack animal cells — including ours — are harder for us to see, because animal cells are reluctant to grow outside of the animal. In 1954, John Enders, Thomas Weller, and Frederick Robbins got the Nobel Prize for developing a method to grow poliovirus on cells in test tubes. A modern epidemiological method is to look for virus DNA in sewage. But the concept is old — as far back as 1932, Philadelphia scientists realized that pandemics in cities could be caught early by detecting polio in sewage. This didn’t catch on, because in 1932, before tissue culture, the status quo for detecting polio was injecting the filtered sewage into a monkey, and then seeing if the monkey caught polio.

Sometimes an innovative idea is not immediately scalable.

But issues of scale aside, the vast majority of microbes haven’t been cultured at all. Even now. Most wild cells just don’t grow in captivity. ⁹ We don’t know why. They might rely on extremely specific nutrition or environmental conditions to grow, or to enter a metabolic state in which they’re willing to grow. Whatever the case, we have resolutely failed to bring the vast majority of life on Earth into the laboratory for convenient study.

There’s more of it

In the 1890s, as early microbiology was trying to make sense of even very basic biochemistry, Sergei Winogradsky, who studied both plants and microbes, was learning that some microbes turn rudimentary chemical compounds into energy. To uncover how soil bacteria turn ammonia into nitrite, and then different bacteria convert nitrite into nitrate, he had to actually culture the organisms in soil. (This work laid the foundation for our understanding of the nitrogen cycle.) He suggested more effort be spent on studying microbes in their natural environments, or at least intentional replications thereof. ¹⁰ Microbes interact with the environment and each other; stripped of context, there’s very little you can glean from one.

And to skip ahead: What we know now is that microbes don’t just interact with the environment; it would be more correct to say they are vast parts of the environment — billions in every gram of soil and every liter of seawater, on and inside every living surface that does not fight tooth and nail to keep them out, making our breathable air, growing our food, tidying up our dead, outweighing and vastly, vastly, vastly outnumbering every other living thing on Earth.

A phrase you may run into is “biological dark matter,” borrowed from the stuff of physics that that the universe seems to be full of, but is invisible and intangible: We can’t culture it, we can barely study it, we know very little about it, but we know it’s there, running the world.

It’s slower

Usually when we study cultured organisms, we study one organism at the peak of its capability: well-fed, replicating fast. But most wild microbial life (which is to say, most life) exists in the lean twilight period known in the lab as “stationary phase” — after the party, after there’s no food left in the system to replicate, after growth stops.

What happens in stationary phase? Well, sometimes, over time, the ecosystem dies. In the lab, this happens if you take a flask of bacteria and leave it — the osmotic pressure, the force driving water out of cells, means that cells are unable to hold together. Nothing survives. In the wild, this can lead to a burst of nutrients released from decaying cells that are then food for new colonists, but it’s not a cycle that can keep going forever. Some cells spill their genes in viruslike packages, so that a surviving neighbor might pick them up and carry them on. Some species produce spores, hardy and inactive, time capsules ready to wake up in a better future.

But the other thing that can happen — in the lab, if you top off the flask with water so the osmotic pressure doesn’t become untenable (but continue to starve them of real nutrients), is that microbes play the long game. They replicate slowly, efficiently, at turnover rate. They grow multiple sets of chromosomes. Their DNA mutates more — at least partly a failure caused by stress-induced damage, but also a fitness advantage. Random mutants, usually at a substantial disadvantage, are more likely to be able to use an existing untapped source of nutrients, or to outcompete their kin if nutrients should come pouring back in. When the going gets tough, the tough get weird. ¹¹

And they wait, alive, for something to happen. They can wait for years.

(The long game, incidentally, might well account for a lot of unculturable organisms. A microbe that reacts quickly to an influx of nutrients is very convenient for human scientists, but that’s just one evolutionary strategy for an event that is, in the wild, quite rare. What’s the rush? Tuberculosis, the bacterial disease whose characterization necessitated inventing the entire field of infectious disease identification, is famously terrible to culture. It’s not especially picky, it just takes weeks to grow. I’m certain that we would never have bothered if we had any other choice.)

There are more weird little guys in heaven and earth

When you start looking even closer, you begin to find that the invisible world is teeming, crawling, swarming with parasites. Where life goes, an entourage of disease invariably follows.

The most important of these are viruses. But there are vast numbers of smaller, stranger replicators that are not viruses but that still go about bouncing from microbial host to microbial host, evolving, moving stray genes about populations, sometimes getting caught in their host’s genomes and sometimes tearing themselves back out. Satellites and virophages realize they can do better than piggybacking on a cell and piggyback on other viruses. There also are viroids, just a loop of parasitic replicating RNA that doesn’t waste time coding for protein. (They reside exclusively in plants, where they occasionally cause disease but mostly get in and get out without fuss.) Fungi have ambiviruses, which have no shell but do encode some of their own replication machinery once inside the cell — part virus, part RNA with a crazy dream. There are prions, replicating elements of pure protein which have no genetic material at all. More like a crystal than an animal, they reproduce by altering compatible proteins into their own shape — and those then go on to alter compatible proteins into their shape. This cascade causes havoc in animals because on the rare occasion they occur, they mostly occur in the brain (and cause, for instance, mad cow disease). But yeasts have several known prion proteins that mostly don’t harm them. They’re very hard to spot but they’re almost certainly out there too, ubiquitous, in the still-dark corners of the invisible world.

Meanwhile, decades of breakthroughs in genetics have led to our best tool yet for seeing the invisible world. Every living thing (aside from prions, if you count them) has a genome: a heart of genetic material, DNA or RNA, that encodes the proteins that make up that creature. Metagenomics is the fine art of stripping the genetic material from many creatures at once — from an ounce of pond water, a gram of dirt or human waste — and processing all their genes en masse.

This capacity is recent. In 1983, biochemist Kary Mullis invented the polymerase chain reaction, a technique which allows scientists to amplify a strand or two of DNA or RNA until they have large enough quantities to easily study. Meanwhile, Norman Pace was hard at work at the University of Indiana, trying to study microbes by extracting nucleic acid samples from the environment. In 1986, Pace and colleagues used PCR to amplify bits of RNA in those environmental samples: Suddenly, it was possible to analyze microbes en masse. ¹² In the mid-2000s, biomolecular innovation and cheap computing began to facilitate the examination of every gene in many critters at once (bacteria, viruses, all of it). You can take a handful of dirt and get a database of the entire genome of every organism. It’s computationally intensive but getting cheaper every year.

This suite of technologies affords us a lot of power. You can compare your list of genomes to databases of known genomes. You can tell how they’re related to other organisms, and each other. You can “search” for known genes by amplifying and sequencing strands of material that contain those genes. So you can identify specific species, yes, but you can also identify if a kind of gene — antibiotic resistance genes, say — is in there, regardless of who has them. Or if you’re interested in, say, all coronaviruses, or all E. coli, you just need a common sequence to search with. Remember the hot new old innovation of detecting viruses in sewage? We got around the monkey!

Metagenomic sequencing is an imperfect tool. It’s kind of expensive. It helps immensely to have a sense of what you’re looking for. It’s easy to lose or discard information from an organism that was very rare in the sample, or didn’t meet the physical criteria you expected. ¹³ But even then, you can get things done by starting wide and narrowing in on weak signals. There are growing techniques for pulling genetic information out of even a single cell. And a genetic overview of even the common organisms in a sample is immensely powerful in giving you a good look at the invisible world.

So having all this genetic information removes the mystery plaguing the culture-based paradigm, right? Well, no! We don’t know what most DNA does.

About 90% of the human genome does not appear to have a clear functional purpose. ¹⁴ That noncoding DNA is an ongoing mystery. Some 45% of it is transposable elements, which are pieces of DNA that copy or cut themselves out of the genome, and then paste themselves back in. One prodigious 300-base-pair-long stretch has pasted itself over a million times in the human genome. ¹⁵ But still plenty is just … there. How did it get there? What’s its deal? We don’t know.

The eukaryotes — plants and animals and fungi and protists — are riddled with nonfunctional DNA. We have it coming out of our gene pockets. Bacteria and viruses, which really benefit from replicating fast and quick, don’t really go for that kind of thing. They use their genes.

But most of those coding proteins are still unidentified! A couple lab-standard model viruses are pretty well characterized, but anything cellular — bacteria, archaea, eukaryotes — is absolutely a toss-up. Strains of E. coli that geneticists have spent careers studying still have mystery genes! ¹⁶ We know almost nothing about thousands and thousands of human genes! ¹⁷ And if you go past the most vanilla of model organisms, the best we can do is compare new genes to the ones in our records … but many, many genes are not clearly analogous to any well-understood ones. There is a great deal of mystery soup. 

Powerful as it is, metagenomic sequencing does not give us the ability to plainly understand what’s going on down there. You have to be smart.

Remember the obelisks? They show us that we can make something of all this darkness by getting clever about it. Usually, metagenomic analysis takes that library of sequenced genetic material and looks for snippets that match up with known sequences. But obelisks had no known sequence. To figure out they were there at all, the team at Stanford abandoned genetic similarity. Taking only unidentified RNA strands, they predicted the physical shapes that they would form. They found short sequences that formed rod shapes, which is on its own unusual for RNA. Looking further at these rod-forming sequences, they found that those sequences: A) were genetically similar to each other, and B) did not have DNA equivalents, suggesting a purely RNA replicator. Finding the obelisks was a matter of taking unidentified strands from RNA metagenomes and asking an abstract question: “What if there was something like a viroid in here? What if it wasn’t related to any known viroid? How would we know it was there?”

The invisible world is still just that. Nobody has seen an obelisk. They were discovered in digital sequence data, an explanation for some of that mystery genetic soup. 

There’s a new conspiracy theory that takes vaccine-denialism a step further: Viruses don’t exist, period. This alternative model makes a number of interesting claims: that nobody has ever seen a virus (false), that that no healthy human or animal has ever been made sick by introducing a virus (false), that so-called viruses are just conglomerations of RNA and protein that are actually released by the human body to fight disease (why do they have similar DNA to each other and not to that found in the human body?), that so-called infectious diseases are caused by toxins (at least it’s not demons).

This alternate model of infectious disease is, charitably, wrong. But having wrestled with the microbial world for a long time, I feel sympathy for its adherents. Scaling any edifice of scientific knowledge requires a lot of faith in other people’s work. Microbiology more than most.

Like: Yeah, we found a virus. It’s in this tube. It looks like it’s water, but I promise, there’s viruses in there. How do you know? We extracted the virus’s vital essence and then ran it through a machine with some other vital essences we made for this purpose, and the machine said it was there. And this makes people sick? Well, we fed some of this water to a rabbit, and the rabbit got sick. Can I see the rabbit? No. You can’t go in the same room as the rabbit. Also, the rabbit died. Can I see the virus, then? No. It’s too small. Can YOU see it? Yeah, we took a picture. Can I take a picture myself? Probably not. Electron microscopy is really hard. Here, look at our photo, though. This is a gray blob. Yeah, we had to take the photo with electrons instead of with light, because the virus was really little and colors don’t exist that small. That cannot be right. Hey, we’re doing our best. Relax. Have a beer.

By the way, you are not going to believe how the beer got like that.