Human Biomolecular Atlas Program Update
By Paul Nicolaus
March 7, 2024 | The NIH Common Fund Human Biomolecular Atlas Program (HuBMAP) is “trying to map—to create a reference atlas—of the 37 trillion human cells in our body,” explained Katy Börner, a professor of intelligent systems engineering and information science and a lead PI on the mapping component for HuBMAP at Indiana University.
She and many other HuBMAP researchers are attempting to better understand the connections that exist among cells within tissues or organs and learn how these relationships impact human health. There is interest in discovering how to “best create a latitude-longitude system for the healthy human body,” Börner told Bio-IT World.
Work is underway on a male and female reference atlas that could help us better understand what is normal when we are healthy and help shed light on how we change as we age or as other bodily changes, such as pregnancy, occur. Rather than mapping a singular human body, however, the underlying idea is to come up with a reference system that works for all. “This is in support of precision health because we want to understand what it means to have healthy bodies, but also, of course, in support of precision medicine,” she said. Similar to the way you can take the human genome and map someone’s genomic sequence to it in order to understand the genetic elements that exist within that person’s genes, it will eventually become possible to use a 3D healthy human reference to learn more about how to live a healthy life and how to treat disease.
Just as it took a great deal of time and effort to arrive at a latitude-longitude system for Earth, “it will take a similar conceptual breakthrough” to come up with a reference system for the healthy human body that applies to the many people living on our planet, Börner pointed out.
Production Phase Underway
Considering the trillions of cells in an adult human body, “determining the function and relationship among these cells” is a massive undertaking, according to a HuBMAP program snapshot. The HuBMAP consortium comprises various research groups from institutions and universities across the US and the world, divided into several areas of focus based on the type of work being carried out by the teams. And while HuBMAP will not necessarily map out the entire body, it is intended to get the ball rolling and put forth a framework and tools that can help enable more complete mapping in the future.
The first phase of HuBMAP involved data generation that included both single-cell and spatial datasets, said Michael Angelo, associate professor of pathology at Stanford University and co-director of the HuBMAP steering committee. These efforts involved building out the database infrastructure needed to not only host those datasets, but also have them linked together through a biologically interpretable ontology that would make it possible to look at a single cell type across multiple organs, for example, or choose the demographic of donors and view all organs.
Essentially, this work meant coming up with a relational database and a blueprint “that brought all those pieces together to where it could be cross-referenced and integrated with all the other mapping consortia,” Angelo told Bio-IT World. He, Börner, and colleagues summarized significant accomplishments made, along with notable challenges encountered, during the first phase of HuBMAP in a Nature Cell Biology paper published last year (DOI: 10.1038/s41556-023-01194-w). The paper also looked ahead to future goals and possibilities.
HuBMAP’s second phase kicked off last year, Angelo explained, and involves three main areas of focus: scaling the generation of 2D data sets, generating representative 3D datasets, and turning all of this data into a resource for the scientific community.
“HuBMAP is currently in the production phase, meaning we have 350 plus experts working within this one project,” added Börner, “and HuBMAP is not alone.” More than a dozen other consortia are also pursuing single-cell studies, maps, and atlases of the human body. “All these different efforts together are trying to improve technologies to get us higher quality, single-cell data.” The initiatives are also setting up portals to enable experts to analyze, visualize, and model the data.
“Every six months, we bring out a new release of the Human Reference Atlas,” Börner explained, which brings together experts across 17 different consortia supported by the National Institutes of Health or the Human Cell Atlas. Nearly 50 monthly calls have taken place to date, enabling experts to present their research, and enabling group discussion on how to arrive at a common coordinate framework. “Ultimately, it’s going to be interesting to see if we can agree on a process to map this data into a common reference system—into the so-called common coordinate framework,” she said. “We all have to agree on how we best combine the evidence that exists to ultimately arrive at something that’s greater than the sum of parts,” she explained. “And that kind of consensus-finding process is nontrivial.”
Börner said one intriguing possibility for arriving at a latitude-longitude system for the human body involves the human vasculature. A short video explaining how the vasculature could be used to create a common framework was published as part of the most recent release of the Human Reference Atlas, she noted.
Emerging Tools and Technologies
Current HuBMAP efforts include the development of new analytical and visualization tools, and Börner highlighted examples of these emerging technologies. A group of papers stemming from HuBMAP research were featured in Nature journals, she said, and as part of that paper package a team from GE helped present 3D data for skin. This work, detailed in a paper co-authored by Börner (DOI: 10.1038/s42003-023-04991-z), focused on understanding what immune cells, damage markers, and proliferation markers are in our skin. She explained that the researchers took samples from several individuals of different ages, with or without UV sun exposure. The 3D data helped showcase what happens to skin as we age or take on more sun exposure. “That kind of data is super valuable for Human Reference Atlas construction and also super valuable for using the vasculature to give us a better understanding of what reference system might work,” she said.
Börner also highlighted a recent Kaggle competition using 3D data on the vasculature.
Participants have embraced this new data, analyzed it, and created machine learning algorithms to segment 3D vasculature in very high-resolution data. Having 3D data is critically important for gaining a better understanding of the body because our bodies don’t operate in 2D tissue sections, she explained. The body works in 3D, and each cell is situated in a 3D environment. Some of those cells are stuck in that environment and cannot move to a new location, whereas other cells travel with the blood or other fluids in our bodies, moving from organ to organ.
And no one organ can operate alone. The brain relies on oxygen from the lungs, for example, and if kidneys fail, other organs are affected. Just as single cells play a role within a larger team, so too, do our organs. The human body is a multiscale, complex system that involves single cells aggregated into larger functional tissue units, Börner explained. Researchers are interested in studying this with greater depth and clarity.
Her team’s work, in particular, has focused on exploring alveoli in the lung, glomeruli in the kidney, and villi in the intestine. “We have now started to do detailed data analysis and visualizations from 22 different functional tissue units,” she said. These tissue units aggregate up to entire anatomical structures, organs, organ systems, and the human body. “We need to understand the coupling of these different levels to each other, and I think data modeling and simulation can be very helpful here.”
She explained that an effort led by one of her postdocs involves work on a virtual reality user interface to communicate different scales, zooming in from the body down to the single cell level, and then to the gene and protein and other biomarker level. There is a lot of data on the cross-anatomical level and a fair amount of data on the microscopic level, but she said more is needed between these two levels.
“The mesoscale level, the functional tissue unit level, that’s really where I think some of the major function is provided that’s needed by all of us in these interlinked organs,” she explained, “but we also need to really understand that level better, so some of the efforts within HuBMAP are focusing in on that.”
Remaining Challenges
One of the main lingering challenges, according to Börner, is that single-cell data remains quite expensive. Because of this, she said it is crucial to utilize data-driven decision-making to determine where to sample next and to figure out which type of assay to apply to any newly cut tissue. “For some anatomical structures, we see data coming to a consensus,” she said. “But then for other areas in the human body, we get datasets coming in, and they all seem to say something different. So maybe we just don’t yet understand the impact of certain donor demographics, or we don’t have the technology yet that is sensitive enough to really disentangle some of the details, or some of the analysis pipelines might need to be optimized.”
Börner also spoke of the need to engage more international researchers moving forward. “We have been inviting experts from developing and developed countries to review the digital objects that make up the Human Reference Atlas,” she said, “and we hope that this also increases the engagement of populations that are currently underrepresented in the experimental data that we see on the different data portals.”
She added: “I think it’s beneficial to all of us to have an atlas that truly represents the diversity we see on this planet.”
Envisioning HuBMAP’s Potential Legacy
“We now have an initial version of the Human Reference Atlas,” Börner said, “so experts have agreed that we can map the human body, and we can also map new tissue to this evolving reference.”
The new releases that emerge every six months may include the addition of new digital objects. There are often new reference organs added because they are “near and dear” to some of the experts joining this effort. Many of the digital objects in the Human Reference Atlas are being adopted by significant data providers, such as the European Bioinformatics Institute in the UK and BioPortal at Stanford University. “And the NIH 3D team at the National Library of Medicine has started to ingest some of the 3D models that we have compiled together with many of the experts around the globe,” she said.
“These data sources are used widely to not just teach but also to run queries against existing ontologies and large databases, so the Human Reference Atlas becomes part of a fabric, part of an ecosystem, of interlinked data sources that are used for research, for training, and also clinical practice,” Börner continued.
When HuBMAP eventually winds down, “it’s very desirable to have the Human Reference Atlas available through data portals that are sustained over the long haul,” she said, and seeing the Human Reference Atlas become part of an interlinked ecosystem is exciting. Ultimately, advancements made during the current phase of HuBMAP could help serve the greater scientific community and lead to future discoveries or breakthroughs in human health and medicine.
“It’s very difficult to understand any disease state unless you really know what normal looks like,” Angelo said. HuBMAP will provide a comprehensive picture of what normal cell states and structures look like at a deep level in all human tissues. This, in turn, will make it easier for researchers to see how diseases veer away from those normal states. Beyond that, he believes the bioinformatics and tools developed along the way will become a notable part of this program’s legacy.
Outreach Efforts
Bringing HuBMAP’s ambitions, initial results, and the “beauty and complexity of our bodies” to a larger audience is an ongoing area of interest, Börner noted. One example of these outreach efforts can be found in the HuBMAP Visible Human MOOC, or massively open online course. Another example is the 24-hour Human Reference Atlas event hosted in 2022. “We are going to host another event just like it but on the multiscale human in December this year,” she said. “I think we all deserve to know more, and we all will benefit to know more about our bodies.”
Paul Nicolaus is a freelance writer specializing in science, nature, and health. Learn more at www.nicolauswriting.com.