sciencebriefs
13:00in productionCh. 1 · A population, not a single organism/ 13:00 · ceiling 15 min
Medicine · Life sciences

Human microbiome

Trillions of microbes live on and in every human body, and genetic sequencing has replaced old guesses about their numbers and roles — but most links between the microbiome and disease remain correlations, not proven causes.

The human microbiome is the community of bacteria, archaea, fungi and viruses that inhabit the human body, and DNA sequencing rather than laboratory culturing has let researchers count and classify it properly for the first time. The Human Microbiome Project's 2012 results replaced rough estimates of microbial abundance with directly measured ones and confirmed genuine roles for gut bacteria in digestion and immune development. What the data has not done is turn most of the microbiome's associations with disease into demonstrated causes.

Chapters & takeaways6
  1. 0:08
    A population, not a single organism

    The microbiome is every bacterium, archaeon, fungus and virus living on and in a human body, popularly treated as an extra organ shaping health on its own.

  2. 2:10
    Counting by gene, not by eye

    Sequencing methods like 16S rRNA profiling and shotgun metagenomics replaced laboratory culturing, which only ever captured species that grow in a dish.

  3. 4:20
    The Human Microbiome Project's tally

    A 2008-launched NIH initiative sequenced thousands of samples from 242 volunteers across fifteen to eighteen body sites, publishing initial results in 2012.

  4. 6:30
    What the numbers actually revised

    The old ten-to-one microbe-to-human-cell ratio was corrected to roughly one-to-one, while real functions like vitamin synthesis and fecal transplant therapy held up.

  5. 8:40
    Where the evidence thins

    Disease associations are mostly correlational, low-biomass samples like blood and placenta are prone to contamination, and no shared 'core' microbiome has been confirmed.

  6. 10:50
    A tool, not a diagnosis

    The sequencing science is solid; the diagnostic and treatment claims built on top of it in commercial microbiome testing mostly are not.

Worth your time?

Selectively. Start with the brief, then study the parts we point at.

4/ 5
What works
  • explains clearly what DNA sequencing measures that culturing could not
  • is honest about the gap between a correlation and a treatment
  • the fecal transplant case shows what a confirmed benefit actually looks like
What does not
  • cannot yet say how much of the disease correlation is causal
  • leaves the 'core microbiome' question genuinely open
Study it if
  • anyone who has taken a probiotic and wondered if it does anything
  • readers curious how claims that people are 'mostly bacteria' were actually tested
  • anyone interested in how sequencing replaced microbiology's old culturing limits
Skip it if
  • anyone wanting a verdict on a specific wellness product or test
  • readers who want the microbiome to already explain mood and weight
The written brief4 min read

A population, not a single organism

The human microbiome is the collective population of bacteria, archaea, fungi and viruses that live on the skin, in the mouth, gut, lungs and other tissues of a healthy person, and the idea attached to it is that this population behaves almost like an extra organ, carrying genes and functions the human genome does not. An early, widely repeated figure held that microbial cells outnumbered human cells by ten to one, used to argue that a person is, numerically, mostly microbial. The claim broadened over the following decade: that this microbial population shapes digestion, immune development, metabolism and even mood, and that disturbances in it, called dysbiosis, contribute to conditions from obesity to depression. The Human Microbiome Project was built to test how much of that picture actually held up under direct measurement rather than assumption.

Counting by gene, not by eye

Rather than growing microbes in a dish, which only captures species that tolerate laboratory conditions, researchers sequence DNA directly from a sample. Targeted sequencing reads the 16S ribosomal RNA gene, present in all bacteria and archaea, to identify which species are present; shotgun metagenomic sequencing reads all the DNA in a sample to see what genes those species carry. The Human Microbiome Project, a five-year National Institutes of Health initiative announced in 2008, applied both methods to more than five thousand samples taken from 242 healthy American volunteers, drawn from fifteen body sites in men and eighteen in women, including the mouth, nose, skin, lower intestine and vagina. Its initial results were published in 2012 across Nature and PLoS journals. A recurring technical difficulty is separating microbial DNA from the much larger amount of human DNA present in any given sample.

The Human Microbiome Project’s tally

Several claims survive the closer measurement. The ten-to-one ratio has been revised down substantially, to something closer to one microbial cell for every human cell by number, though microbial genes still outnumber human genes roughly a hundredfold, at about two million against twenty thousand. Gut bacteria, dominated by two bacterial phyla, do ferment dietary fibre into short-chain fatty acids and synthesise some B vitamins and vitamin K, functions the human body cannot perform unassisted. The immune system genuinely interacts with resident bacteria during infancy, when colonisation appears to programme lasting immune responses. And one clinical application is well established: faecal microbiota transplant, which introduces a donor’s microbial community into a patient’s gut, treats recurrent Clostridioides difficile infection with an effectiveness researchers put at roughly 85 to 90 percent, a rare case where manipulating the microbiome demonstrably changes an outcome.

What the numbers actually revised

Much of the rest is less settled than headlines suggest. Associations between microbiome composition and conditions including type 2 diabetes, obesity, inflammatory bowel disease, Parkinson’s disease and depression are real correlations in the data, but the studies establishing them mostly cannot show which way causation runs, or whether some third factor drives both the microbial shift and the disease. Claims for a distinct microbiome in the placenta or in blood remain disputed for a more basic reason: at very low microbial biomass, the sequencing process itself is prone to picking up contamination from reagents and lab handling, and researchers cannot yet reliably separate a genuine signal from that noise. There is also no confirmed ‘core’ microbiome shared by all healthy people; composition varies enormously between individuals and even within the same person over time, which undercuts the idea of a single normal microbial baseline to compare against.

Where the evidence thins

The practical stakes are already visible outside the sequencing lab. Faecal transplant is a real treatment used today, and how an infant is born measurably affects its early microbiota: babies delivered vaginally pick up maternal bacteria at birth, while those delivered by caesarean section carry more of the bacteria common in hospital environments and take longer to acquire the community associated with vaginally born infants. Large public sequencing efforts, including the Earth Microbiome Project’s attempt to catalogue microbial diversity across environments generally, extend the same methods well beyond human health. There is also a less comfortable consequence: a person’s microbial DNA is distinctive enough to help identify them, which means microbiome data collected for research carries a privacy risk resembling that of genetic data, even when it is shared anonymously.

A tool, not a diagnosis

This is worth understanding precisely because the underlying science is more interesting, and more limited, than the wellness-industry version of it. The sequencing work genuinely replaced guesswork with counted data, corrected an inflated ratio, and identified real microbial contributions to digestion and immune development. It has not yet delivered the causal story that popular coverage often implies, where an unbalanced gut explains mood, weight or chronic disease; most of that remains correlation awaiting a mechanism. Readers deciding whether a probiotic supplement or a microbiome test is worth acting on should treat the diagnosis end of that industry with more scepticism than the underlying biology deserves, since the biology itself is well supported but the diagnostic and therapeutic claims built on top of it mostly are not.

Same field · Medicine4 of 88
Up next in Science

RNA interference

· 13:00

In 1998 Andrew Fire and Craig Mello found that double-stranded RNA, not single-stranded RNA, silences genes with striking potency in roundworms, revealing RNA interference as a cellular pathway now used as both a lab tool and a medicine.

13:00