How can immune repertoires capture the imagination of the public like the microbiome has?


I imagine that everyone is aware, the microbiome has been and continues to be a “hot” topic. In the United States, the human microbiome project (HMP) launched in 2008 with millions of dollars of funding, and just recently there was an announcement from the White House for a call to action on microbiome research.

If you are reading this, you know that repertoires both determine our resistance to infection and are diagnostic of past exposures and current immune system problems. Personally, I would say that repertoires are at least as predictive and diagnostic as the microbiome. We aren’t yet at the point when we can easily make inferences of using repertoire sequences, but the microbiome folks are also just in the beginning phases of understanding how to connect the microbiome with health states (with a few notable exceptions).

So, how can we capture the imagination of the public? This image contains most of the elements that I think have made microbiome analysis so popular:

  1. This is from a citizen science project, in which you send your poo and they send you a report. A number of startups have done the same thing.
  2. Broad sampling and public sharing of data in an organized fashion. There were whole segments of the HMP that were dedicated to data organization.
  3. Easy visualizations. The principal component plot of microbial occurrence, while it has its limitations, gives a very nice “you are here” summary.

What sorts of analogies could we make with repertoire analysis? More generally, how can we get people excited about repertoires?


the difficulty is that a whole bunch of bugs in your gut is a lot easier to think about than a whole bunch of sequences of b-cell and t-cell molecules in your blood. so, maybe some graphics would help?

and somehow tie in to how each individual’s immune repertoire is distinct, changes day to day and after cancer treatment, and how following the repertoire sometimes helps personalize cancer treatment. The new results from cancer immunotherapy for brain melanomas, and lung cancer etc are starting to show up in the popular press a lot…


Done and DONE.

Okay, not quite, but here are some potential avenues for the future (my bias is B cells):

  • Clones as stamps of past infections. We each have a unique fingerprint of past antigen exposures, which we can eventually link with high confidence to different strains. It would be cool to see the densities of clones to different things rise and fall and get committed to memory.
  • B cells as indicators of recent exposures (ooh, at some point in the past year or two, my old anti-RSV lineages evolved a bit further!)
  • Personal evolutionary dynamics… each tree is like a scrapbook of evolution. Maybe with the right animations, this will help bring home evolution more generally.
  • Personalized risk assessments. For instance, my repertoire is skewed toward epitope A of H3N2, so maybe I’m protected this year.
  • (The complement) B cell repertoires as indicators of gaps in protection. We can watch vaccines fill them.
  • Better understanding of autoimmune diseases.

Obviously, some of this can be done with serology, but B cells make it vivid and personal. The problem will be that the samples are relatively sparse (due to limits of blood draws), so good inference is vital.


We could generate similar reports to the one shown above. I think eye-catching visualisations are the key, even if we don’t fully know the implications of different repertoire features at this point. A few ideas below, and what would be useful to do to help achieve them.

  1. Your repertoire is diverse! I find that people’s eyes are always drawn to a nice network representation of a repertoire. Even if it doesn’t say much personal, it is something that looks fun, and illustrates how diverse the repertoire is.
  2. How does your repertoire compare to others? PCA of broad repertoire characteristics and where you fit (diversity, CDR3 length, mutation, VJ, etc). Need a good baseline dataset of healthy people so we can then see how people compare to this.
  3. How mature is your repertoire? We know how some repertoire features change with age, and would be fun to see if we could predict the age of the people based on their repertoire features alone.
  4. What pathogens have you been exposed to? This is probably the most interesting, but also the most difficult at the moment. Relies on having a database of previously described antigen-specific sequences with which to annotate the datasets (see this thread). While currently the number of sequences of known specificity is fairly limited, I think this will rapidly grow as more people apply RepSeq to various disease/vaccination cohorts, and we still may be able to get a rough idea.

Does anyone know of any immune repertoire startups already doing this kind of thing? If we could get people interested in having their immune repertoires sequenced, it would then be a great research resource too.


This is a key point - though I’m basically a layman with respect to the gut microbiome, my impression is that they aren’t that much ahead of “us” in understanding the implications of different features.


My sad analogy:

Microbiome data - ~1000 OTU, ~10^6 “16S RNA” records in GenBank + phylogeny and decades of microbiology studies
RepSeq data - 10^6 clonotypes (with the number of unique single chain TCR amino acid sequence variants > 10^8) and around a thousand records in GenBank.

The only figure I’ve created so far that in my humble opinion can be catchy for the public (after proper formatting of course) is this one:

It basically shows how our repertoires diverges with aging because of various factors (mostly antigen-driven expansions). Those are unrelated donors, “0” points mark the cord blood. The closer the points are the more T-cell clonotypes overlap between them.


I came across the Human Vaccines Project recently: I wonder if anyone here is involved in it (paper below). I think it’s the kind of initiative that could capture imagination - and might also be something we could hook into.


Yes, our laboratory is leading the Human Immunome Program in the Human Vaccines Project network. We are slated to sequence complete BCR and TCR repertoires for 1,000 people, and have already started the process. The goal is to identify all of the expressed BCR and TCR on the planet. We have spent the last year interacting with major donors to finance the longer term aspects of the project. In general, I have found that members of the public (and the scientific community for that matter) don’t have a vision for how the complete Human Immunome might impact their lives or work. To me it is obvious that Human Genome >> Precision Medicine and other initiatives, and Human Immunome will >> same magnitude of impact. But most people want to know the more immediate impact for them (“will this inform cancer immunotherapy” “how will this impact MS, or IBD, or allergy , etc etc”. It is apparent we will need to deliver some early wins in comparing disease state immunizes with healthy for the public to see the potential.

I did a talk Unraveling the Mystery of Immunity at Tedx recently to try to articulate some ideas around Human Immunome, although it was sort of pitched in the context of infectious diseases and vaccines, and I described but did not explicitly name the Human Immunome Program work.


Wow, I hadn’t seen that project, that’s neat. Having 1000 [BT]CR repertoires along with demographic data would be wildly useful. Casual googling sounds like this is 10 years out… is the best way to stay updated to just keep an eye out for when you publish papers, or what?

And to keep from drifting the thread too much, I agree, if we want to get the microbiome people’s grant money, we should probably make more colorful networkey plots that we can’t really interpret meaningfully. And probably talk about poop more.