Human Cell Atlas

Cori Bargmann discusses Chan Zuckerberg Initiative’s Cell Atlas program at Wellcome Trust.

Cori Bargmann:  In early 2013, I was asked to serve on an NIH advisory committee to plan President Obama’s BRAIN Initiative. This is a systems neuroscience project to learn how the brain uses the chemical and electrical activity of millions of neurons to generate perception, memory, and emotion in real time, to “construct a dynamic picture of brain function that integrates neuronal and circuit activity over time and space.” The committee studied the field deeply and emerged with seven initial recommendations almost exactly three years ago. The first recommendation was to define and describe all of the different kinds of neurons in the brain. The goal was turned into an NIH request for applications that followed, and led to some of the initial single-cell RNA sequencing projects described earlier today by Aviv Regev and Arnold Kriegstein. These are some of the early steps toward the Human Cell Atlas project.

The striking thing about this story is that there is no field of mechanistic biology less likely to reach the conclusion that a cell atlas is important than that represented by the BRAIN Initiative. This recommendation emerged from a field steeped in electrophysiology, a field that studies signaling across extensive networks over milliseconds, not a molecular field. Neuroscience was the exact opposite of immunology, mentioned earlier today by Christophe Benoist, which was “ready” for a cell atlas because it had historically taken cell type classification as the central construct, and viewed transcription as a key output. Not the case for neuroscience. But nonetheless the idea emerged from this field as well. It’s a sign that all of biology is ready for this approach, and that the Human Cell Atlas is a project whose time has come.

I am no longer immersed in the BRAIN Initiative; I have a new identity now. Starting next week I will be leading the Chan Zuckerberg Science Initiative.

Last year, on the birth of their daughter, Priscilla Chan and Mark Zuckerberg pledged to give away 99% of their wealth to advance human potential and promote equality. Last month, as part of this pledge they announced Chan Zuckerberg Science. It has the goal of supporting and advancing science and technology that will allow us to cure, prevent, and manage all disease by the end of the century.

Let’s pause. The goal is ambitious. Could it be realistic? 80 years is a long way away. 80 years ago, the first antibiotic was developed, and as a result, none of the four most common causes of death in 1900 – all infectious diseases – are in the top ten causes today. (ref demography.cpc.unc.edu/2014/06/16/mortality-and-cause-of-death-1900-v-2010/ figure, original citation Centers for Disease Control). Tremendous advances have been made, and the next 80 years should see even more. Now is the time to be ambitious.

If we take this goal seriously, how do we proceed? Mark and Priscilla, and I, have spoken to many scientists and physician-scientists, and engaged formal and informal advisors in this discussion, including people in this room. We have a plan with three elements. First, the unsolved problems in biomedicine are complex, and require interdisciplinary solutions. We think the best approach is to bring together expertise in different areas to solve them. That includes basic scientists and clinician-scientists, and in the modern era it also includes computational biologists and engineers, because these are powerful, fast-moving areas with great potential that are still barely incorporated into biology. We want to bring these people together, and create new scientific models in which they have incentives and rewards for working together on difficult problems over the long term. Second, historically we know that the greatest advances in biology have come from new tools, at least as much as from new ideas. The cell theory emerged from the microscope. Tools and technologies empower all scientists and elevate all of science; if we want to accelerate the whole field and all scientists, this is the place to start. Third, no philanthropist, no matter how generous, can solve all diseases alone. The last arm of our plan is to build support and funding for science from every quarter – public, private, and philanthropic. We consider this outreach a central part of our mission, a project to bring the scale of science up to the scale of the need and opportunity.

To implement the scientific goals of Chan Zuckerberg science, we are initiating three research programs. The first program, and our first investment, is the Biohub. This collaborative new research organization combines Stanford, UCSF, and Berkeley scientists with Chan Zuckerberg scientists to model the new scientific interdisciplinarity, and serve as a prototype of a new kind of science. Steve Quake, the Co-President of the Biohub, will tell you more about that later today. The second program is the development and support of transformative technologies that will advance all of science. The third program is problem-based cooperative networks of investigators that will work in challenging scientific areas. We are still developing all of these programs, and will have more to say about them in the future. But today: We view the Human Cell Atlas as a Transformative Technology – a resource that will advance our understanding of all human diseases. One of the topics in the room is whether this is the time for a Human Cell Atlas; at Chan Zuckerberg Science we believe that it is, and we are committed to it and want to support it in partnership with the people in this room and others around the world.

Partnership is key. At the Biohub, our partners are UCSF, Stanford, and Berkeley. In the Transformative Technology projects and Challenge Networks, we will need partners. Our partners will be scientists around the country and ultimately around the world, and the universities and institutions at which they work. We hope that they include many of you here today, and we want to approach you with humility as we plan what to do. Our partners will include other funding organizations, like the NIH and the Wellcome Trust who are also moving toward the Human Cell Atlas as a goal – I’ve already talked with several people in this room to help refine our piece of the puzzle. At Chan Zuckerberg Science we are not competing with NIH or duplicating it, and we are not of its scale and scope! But we believe that with our emphasis on computation, engineering, and technology, we will have a different center of gravity that will complement the NIH’s efforts and Wellcome’s efforts, and those of others to come.

In discussions with our advisors and many of you, we have identified certain areas where we would like to move this field forward. In the immediate term, the Human Cell Atlas needs a foundation, with certain topics that have been touched on by others here today as well. For the RNA sequencing projects, we need to optimize robust preparation of samples – with difficulty ranging from easy, like blood, to really difficult, like brain and pancreas. Do we want fresh, frozen, fixed, nuclei? How do these samples compare to each other? We need to compare different RNA sequencing approaches, broad, deep, short-read, long-read. We need to address variability between individuals and labs. And we need extensive computational development and suites of tools to collect, visualize, analyze, and share the data. These problems are not glamorous, but they are critical for progress! We want to enable work to solve them quickly and collaboratively, and share progress as quickly as possible between groups so that the whole field progresses.

In the longer term, because we have a long time horizon, we want to look past the immediate technology of single-cell sequencing to longer-term goals. Most of the projects we’re discussing now are really a cell census – a list of the types and numbers of cells, and their approximate locations – and they’re based on RNA sequencing of dissociated cells. But a true cell atlas will be a map that arranges this information across scales from molecules within cells, to relationships between cells in the body. As a neuroscientist, I know with every bone in my body that 86 billion dissociated neurons are not a brain. Technologies that are currently in development will move toward that goal of a true atlas, looking at cells in the animal. One such technology has been alluded to several times already, a method for mapping many hundreds of RNAs back onto tissue slices. People working in this area include Xiaowei Zhuang, Long Cai, George Church, and others. But we want to think bigger, too. Something that Fiona Watt mentioned briefly are new tissue clearing methods like Clarity and iDisco, developed by Karl Deisseroth and Marc Tessier-Lavigne’s labs, that make it possible to visualize molecules in an entire unsliced organ. We can already see the entire sensory nervous system of a mouse in a single transparent mouse, we can see all of the cells in an adult kidney in a single transparent kidney. Other new methods like expansion and compression microscopy, developed by Ed Boyden and Ali Erturk’s labs, can make it possible to see these tissues on different spatial scales. This is an area that we want to support and develop at Chan Zuckerberg Science to move from a cell census to a cell atlas. Can we make these methods 1000 times cheaper and 1000 times faster? If so, we might do the whole atlas this way. And even if this takes longer, it might be that technologies for whole-tissue atlas will develop from the initial Human Cell Atlas project, just as technologies for next-generation sequencing were developed – but not yet ready for use — in the Human Genome Project.

We’ve discussed the computational challenges of single-cell RNA sequencing many times today, but the computational and engineering challenges of a Human Cell Atlas that is mapped across tissues, as suggested here, will be immensely greater and more complicated. We need to develop those approaches in common, from the outset, to move toward our goals. The Allen Brain Atlas has taken impressive steps in this problem already, and at Chan Zuckerberg Science we hope they’ll serve as an example and partner in thinking through the next generation of solutions.

In both short- and long-term goals of Chan Zuckerberg Science, we’ll need to use and deploy an extensive computational and engineering infrastructure. It will need to be built, and maintained, and validated. This is sometimes a secondary part of scientific projects; we view it as a primary priority and want to support it for the community, in partnership with many of the people who are here. We view technology dissemination the same way, and agree with the principle of sharing and disseminating data as a core element of this large-scale venture.

Many aspects of the Human Cell Atlas that we support echo principles that have been stated by many in this room. Repetition is tedious, but here repetition shows us that there is a consensus that we are reaching for the Human Cell Atlas, and that is no small thing at this early stage.

At Chan Zuckerberg Science have a mission focus, and will support research on the Human Cell Atlas consistent with that mission. It’s important to us that we accelerate progress – that the data be shared (very quickly, within the pilot projects; in common repositories that are open to analysis in the later stages), and that the best practices are developed and moved out into the community quickly. It’s important that the project be comprehensive across all normal tissues; our goal is to address all diseases. We have a strong bias that the best projects will incorporate both experimental and computational elements, but we are really looking to this group and others in this field to come up with the best ideas for solving these questions. We are at ground zero at Chan Zuckerberg Science in terms of logistics, but we are enthusiastic about getting started! And in addition to supporting the community of scientists represented at this meeting, we will support the Human Cell Atlas through work at the Biohub that Steve Quake will describe to you.