Skip to main content
Premium Trial:

Request an Annual Quote

Hiroaki Kitano Discusses Systems Biology’s Moonshot: A ‘Virtual Human’ in 30 Years

A collection of systems biology researchers has called for a large-scale international project to create a detailed computational model of human physiology, or a “virtual human,” within the next 30 years.
In a statement issued at the recent Future Challenges in Systems Biology conference in Tokyo, representatives from several international research groups recommended that the systems biology community coordinate its efforts in order to achieve this ambitious goal.
The statement, called the Tokyo Declaration, notes that “the time is now ripe to initiate a grand challenge project to create over the next thirty years a comprehensive, molecules-based, multi-scale, computational model of the human (‘the virtual human’), capable of simulating and predicting, with a reasonable degree of accuracy, the consequences of most of the perturbations that are relevant to healthcare.”
This week, BioInform spoke to Hiroaki Kitano, director of the Systems Biology Institute in Tokyo and an organizer of the conference, about the next steps in achieving this goal. The following is an edited version of the conversation.

Can you tell me how the Tokyo Declaration came about? Why did you decide to formalize this goal of a virtual human in a statement, and what do you expect to achieve through this step?
The background is that the [UK’s Biotechnology and Biological Sciences Research Council] and the [Japan Science and Technology Agency] decided to organize a joint workshop on systems biology in Tokyo [earlier this month], Future Challenges for Systems Biology, … and we decided to focus this workshop on the possible future paradigm of systems biology and medicine, and in particular on the drug-discovery aspect. 
After three days of discussions, we decided that we need to have a long-term vision on how we should proceed in systems biology with regard to how we can contribute to drug discovery and medical research. And there were a few options [proposed], but one option was to create a fairly comprehensive molecular and mechanically based simulation model, which we termed the virtual human. 
So that was kind of the consensus, and it’s been argued that this would be an interesting project to launch, but it’s been relatively unclear how and when we should proceed. But, we start understanding how to do it and technologies are being developed to make it happen. Thus, during the workshop we came to the conclusion that although it’s challenging, we should seriously talk about starting this kind of project, and that’s why we decided to make a statement, that it’s time to go ahead with this kind of ambitious project. It’s difficult and it’s challenging, but I think at least we should start talking about it in a serious way, and that’s what we agreed — that we should issue a statement, and over the discussion it converged into the Tokyo Declaration.
Can you clarify what you mean by ‘virtual human’? Would this be a molecular-scale model for all human cell and tissue types?
We’d probably have to start at the cellular level, and [in some cases] at the organ level. For example, there are already a few projects — some in the UK, some in Japan — [in which] people are bringing together the molecules, the ion-channel mechanisms, the cardiomyocytes for a heart model, for example. Dennis Nobel at Oxford [University] has this kind of model, and also [Akinori] Noma at Kyoto University, but it’s kind of fragmented at this moment, so I think it’s a good time to start with a fairly detailed model based on all the molecular interactions inside the cell, and scaling up to the organs, and then the entire human system.
Of course, we can’t really do everything at once, but we can start with the cell and then bring the capabilities to model various aspects of cellular systems and various different cell types, and then build that up to an organ model, like the heart and probably the liver. Germany is actually spending quite a bit of effort on the HepatoSys liver-simulation project, and probably we can come up with some other interesting projects to properly integrate cellular-level simulation models with organ-level simulation models.
And at the end we actually need to bring them together to create the whole-body model. In terms of drug discovery, you want to have an ADME-tox model, and [for that], you’d require the entire body circulation, how the drug diffuses into each organ, and how that impacts specific molecular interactions. So that’s a grand goal, but we are actually starting with something more practical, obviously, talking about specific cell types and specific organ systems.
One of the discussions we had was how the space program was planned. People didn’t try to go to the moon from the beginning. For the US program, there was the Gemini project, which was actually an earth-orbiting project, and then there was the first phase of the Apollo [Program], which was just circling the moon instead of landing on it. And then they started the moon landings.
So I think we need to have some stepwise phases, for example, prior to the human cellular level and the partial organ-level model, we specifically stated that we need to have microbial modeling as well. Probably yeast and one or two pathogens, and that would be very important because yeast, for example, is something we really know in detail. The entire sequence is known and most of the interactions are sort of known, so it’s really an ideal model for a pilot study.
And we can also do something similar for some of the pathogens, because one of the possible targets would be infectious disease, for example, and then you would need to have interactions between the pathogens and human tissue. You can’t really cure infectious disease only by studying human beings. You have to study the pathogens and the human being, so it really makes sense to have a relatively simple — but still very difficult — unicellular system.  
Where did the proposed time frame of 30 years come from?
[Laughs.] That’s kind of debatable. I said that it would be 2050, but some people wanted to make it a little more aggressive. Before moving into the systems biology field, I started a robotics competition called RoboCup, and the goal is to have a team of fully humanoid autonomous robots beat the world champions in soccer by 2050. So I thought that 2050 would be a good target for this project as well, and by 2050 I’ll be around 90 or so.
But 50 years, for some people, is a little bit too long, and 30 years is one generation, so that’s a good timeframe. It’s very challenging to do in 30 years, but I think it’s good to have an ambitious timeframe.
You mentioned that this would need to be done in a stepwise fashion. Do you have any concrete goals over the next five to 10 years that you would like to see accomplished in order to lay a foundation for this?
We haven’t discussed too much in terms of specific milestones. We know in the next five years we’ll have to create more integrated models of relatively simple organisms like yeast or some pathogens, and we’ll have much better understanding of specific types of mammalian cells. For mammalian cells, it’s probably too complicated to have a comprehensive cellular model in five years, but [we might have] a broad-coverage model in five years. I don’t know how accurate it will be, particularly for the mammalian cells, but we should be able to have reasonably good accuracy in part of the cellular system, and that can be improved over years. So we’ll see how things go.
We’re planning to have a series of follow-up meetings to actually hammer out the milestones, and then we’ll have a better answer about whether this 30-year timeframe is actually the proper one, or if we’re too ambitious, or if we can go much faster. I don’t expect that we can go much faster than that, but whether we can stick to the 30-year timeframe, or we have it done by 2050, we’ll see by a series of follow-up workshops we are planning right now.
What do you envision as the primary challenges in this? Is it the lack of data to build accurate models, or standardization, or computational capability?
Actually, everything. We don’t have sufficient data and sufficient accuracy and comprehensiveness. And how we actually compute that — people use [ordinary differential equations] or stochastic simulation, but we have to have the spatial aspect as well, and if we have a realistic and very large model, stochastic simulation is just way too expensive in terms of computational costs, but the question is what is the alternative. So how we actually use everything we have is the issue.
And also, the intracellular structure is not just packed with all the molecules — it’s a very specific, fine-grained heterogeneous structure inside the cell, and we won’t be able to simulate that in a precise sense right now. So we have to find out how to compute so that that’s a more algorithmic or theoretical aspect of simulation as well.
So actually, everything has to fit together. I think we’re making progress in every aspect, but it’s not enough. We have to do more, and we have to do it with a specific target and a specific goal, and that’s what this statement is about — that we actually put our heads together to actually delineate what are the missing components as well.
Do you have an estimate of what sort of funding this would require? Have you made any cost projections?
Well, this will be big money, obviously, and that actually requires multinational collaboration. I don’t think any one country would be able to put together all the expertise and funding to accomplish this.
I don’t know the exact amount of funding that this would require, but probably substantial funding, a few hundred million or probably more over a 20- to 30-year period of time.
But I think we should have a more stepwise [approach], because if you spend too much money in the initial phase, there will be a lot of hype … and too many expectations. And of course, we’re talking about a 30-year project, and that means, actually that in the next five to 10 years we won’t see the physical result in terms of satisfying expectations.
We’ll have to carefully coordinate this. Actually, one of the reasons we’re saying this will take 30 years is to signify that this is a challenging project, so don’t expect that we’re going to have an immediate return or a five-year return. So I think it’s important that we have scalable funding in a way, and we’ll also need the funding over a relatively long term.
We really just made a statement of where we want to go. How we want to go, and how we get funding for that, is the next thing we have to think about.

Filed under

The Scan

Long COVID-19 Susceptibility Clues Contained in Blood Plasma Proteome

A longitudinal study in eBioMedicine found weeks-long blood plasma proteome shifts after SARS-CoV-2 infection, along with proteomic signatures that appeared to coincide with long Covid risk.

Tibetan Study Finds Adaptive Variant Influencing Skin Pigmentation

With a combination of phenotyping and genetic data, researchers document at PNAS a Tibetan-enriched enhancer variant influencing melanin synthesis and ultraviolet light response.

Domestication Linked to Nervous System Genes in Inbred Mouse Strains

Researchers highlighted more than 300 positively selected genes in domesticated mice, including genes linked to nervous system function or behavior in Genome Biology.

ALS Genetic Testing May Be Informative Across Age Ranges, Study Finds

Researchers in the journal Brain identified clinically actionable variants in a significant subset of older ALS patients, prompting them to point to the potential benefits of broader test use.