Skip to Main Content

Everyone, it seems, uses Facebook and LinkedIn these days. Social networking sites make it incredibly easy to find information about people. You can look up new acquaintances, professional contacts, even the author of this article to figure out who is worth connecting to and who might be trouble.

As a systems immunologist, I wish we had a Facebook for cells. Say you’ve encountered Dan in a tissue sample from a patient. He’s a member of a group of like-minded dendritic cells and lives in the skin. Dan hangs out with Tess and Bess, from the lymphocyte group. Tess talks to Bess a lot, and every now and then gets her riled up. When this happens, the patient feels worse.

That is, in essence, the social media story of atopic dermatitis. If we can detect and disrupt the conversation between Tess and Bess, we can make patients feel better.


Picture the immune system as a society of cells, with many of the same characteristics that make up modern human society: diverse, interconnected, and dynamic, characterized by near-constant communication. Think Facebook, only at a greater magnitude and more complex.

Immune-mediated diseases arise when cell-to-cell communication goes haywire. But if we can change how cells talk to one another, we can ease disease. Figure out who the troublemakers are and we can cure. But with 37 trillion cells in the body, humans can’t make sense of all of those conversations without some help.


The first step in understanding the complex cell communication system is to measure it. High-throughput single-cell genomics protocols such as the one I helped develop at Harvard Medical School and widely adopted commercial solutions have lifted the curtain on the social network of cells in health and disease, helping us understand how immune cells talk to each other on a large scale.

Single-cell sequencing works like this: RNA from millions of individual cells can be measured and analyzed, creating a distinct RNA fingerprint for each cell type. That fingerprint can be used to recognize cells and monitor changes to their abundance and activity. My team at Sanofi — along with academic and industry partners around the world — is now using these protocols to uncover the essentials of cell identity: their origin, function, and interactions. Instead of simply hearing what is being said, we can observe who is talking and who is listening.

For people with autoimmune conditions, these conversations define the difference between health and disease. Understanding which cells are saying what gives us critical information on how to silence damaging conversations between cells that make someone sick while letting the healthful ones continue.

New technologies, however, often present new challenges. Reconstructing cellular profiles does not automatically expose who the troublemakers are. Think about doing the same on social networking sites. It’s no easy feat: human analysts are limited in their processing capacity and are often hampered by subjective biases. This is why Facebook, Twitter, and other social media sites increasingly rely on artificial intelligence to detect, expose, and prevent malicious activities.

With the emerging Facebook of cells, state-of-the-art analytical tools have been largely dependent on scientists to interpret cellular profiles. This creates a bottleneck that hampers the wider adoption of single-cell genomics in the biopharmaceutical industry, which is why we at Sanofi, and others around the world, are developing AI algorithms as a natural, scalable solution to recognize and monitor cell identities.

In this context, AI is no longer a luxury but a necessary component of a new technology. To understand a complex system, it is not enough to invent a way to take more measurements. We must also invent algorithms to make sense of these measurements in an unbiased, reproducible, and scalable manner. My colleagues and I use these advanced analytics to inform our novel target programs and understand which patients are more likely to respond to a novel treatment.

Thanks to this growing library of algorithms, it is possible to imagine a world of scientific discovery in which every biologist is privy to the Facebook of cells. By using AI to monitor cell interactions, we can target known troublemakers and ultimately take control of the conversation to improve outcomes for our number one focus: patients.

Virginia Savova, Ph.D., is principal senior scientist at Sanofi U.S., based in Cambridge, Mass.

  • Considering a Facebook-type social network for “cell communication” seems a very far-fetched strange idea. Firstly because FB’s side-effects have shown to be very damaging (now under US court scrutiny). Secondly because cells do not know the alphabet and can’t type. This article just seems odd sensationalism (paid by FB?)

Comments are closed.