Interview with Daniel MacArthur

This is an interview with Daniel MacArthur, whom I first met when he was at Harvard Medical School, Massachusetts General Hospital, and the Broad Institute of MIT and Harvard in Boston. During his time in the US, Daniel co-directed both the Medical and Population Genetics Program and the Center for Mendelian Genomics at the Broad Institute. He has since moved back to his homeland of Australia.

In a layperson’s terms, what is your current research agenda?

I direct the Centre for Population Genomics in Australia, which is a joint effort between the Garvan Institute of Medical Research and the Murdoch Children’s Research Institute. Our focus is on building a more equitable approach to genomic medicine, which is a collective term for a broad set of technologies designed to improve the prediction, diagnosis, and treatment of disease. Our overall goal is to ensure these technologies are applicable to everyone, not just a privileged few.

The Centre has three major programs. Firstly, we use genomics to understand the patterns of genetic variation in under-represented communities - our program OurDNA focuses on engaging with people of Oceanian, Southeast Asian, Middle Eastern, and East African ancestry. We’ve invested a lot of time building strong relationships with these communities in Australia, and phase 1 of the OurDNA program will be recruiting 10,000 individuals from these backgrounds and collecting DNA, live blood cells, and other health information, to understand how DNA affects health. We are also collaborating with national Indigenous-led efforts working to empower Aboriginal and Torres Strait Islander people in genomic medicine and research.

Secondly, we work on bringing together two powerful genomic technologies - whole-genome sequencing and single-cell RNA sequencing - at large scale, to understand how genetic variation impacts the biology of the blood and immune system, as a way of looking for new therapeutics. And our final program uses large-scale genomic technologies - and more recently, AI and machine learning - to find the genetic changes that are responsible for severe genetic diseases, and ultimately to help design better therapies for these conditions.

Describe your experience with funders (private, NIH, others). What do funders do right? And what could they improve?

I spent the first nine years of my independent research career within the NIH research system in the US, and then moved back to Australia in early 2020, so I’ve now had a fair bit of experience working with the Australian funding system as well. In both the US and Australia a substantial chunk of my team’s research funding has come from industry and from philanthropy, as well from government-funded grants.

Regarding federal funding, the process is very different between the US and Australia, and not just in terms of total funding. One thing that struck me is that NIH is much more strategic than the Australian system: at NIH, there is a greater degree of scientific expertise among people defining funding priorities (e.g., program officers), and the major programs they launch are generally pretty well-aligned with where the field is heading. Then there’s also the review process at NIH: it’s not perfect, but because it has a large pool of expert reviewers to draw on, decisions on grant awards are generally better calibrated than in the Australian system.

One thing funders in all countries struggle with is incentivising consortium research. Many fields, like genomics, require large-scale collaboration to make progress, because that’s the only way to build the big resources that individual scientific teams can then use to tease out discoveries. Funders often do this by funding multiple individual teams, but then either leave them to their own devices, or jam them together into top-down consortium structures that don’t make a huge amount of sense. The most functional consortia I’ve ever been part of are the ones that formed organically, cobbled together early-stage funding, and then got large-scale funding once the team and ideas had solidified - but this is actually really hard to get support for in most current funding systems.

Most scientists with ambitions beyond individual projects end up cobbling together funding support from multiple sources: philanthropy is critical for more ambitious and transformative work; government grants provide support for less risky projects; and I’ve found industry-sponsored funding to be incredibly important for anything with a clear arc towards clinical translation. So there are ways to get big things done in the current funding system, but it’s almost always a struggle.

Surveys show that scientists say they spend upwards of 44% of their time on proposals, reports, IRBs, budgets--that is, bureaucracy. Is that consistent with your experience? Is there anything that could be streamlined?

I think that’s a pretty universal experience, and it’s something that’s driven by both funders and research organizations themselves.

Some of this burden is unavoidable: the world continues to increase in complexity, and maintaining compliance with an ever-increasing set of regulatory rules is hard. It’s also true that any scientist who wants to do something really ambitious needs to accept that a lot of their time will be spent on things that aren’t science - management, collaboration, and running an organization of any scale takes real work.

But I think the sheer degree of complexity we now face is also a consequence of poor incentives within organizations. Any organization that survives for long enough inevitably accretes more bureaucratic processes as it becomes larger and older. This is a kind of organizational senescence, and can only be pushed back by relentless pressure from leadership to reduce unnecessary processes. The organizations I’ve seen that work well have leaders who are really focused on constantly stripping away things that aren’t entirely necessary for the science to get done. And that’s pretty exhausting - most academic and funding leaders don’t want to spend time in the weeds like this.

One major challenge that I think we don’t talk about enough is that so many of the organizations we spend time in are large, ancient institutions that are not actually designed for research. For instance, people strongly associate research with universities, but if you were to build an organization from scratch to efficiently facilitate research it would be very unlikely to resemble a modern university. The same is also true for hospitals. These are organizations for whom research is important, but their core systems are fundamentally designed for other purposes that provide more of their revenue: for teaching, or treating patients, for instance.

I don’t blame individual administrators, who are generally doing their best in a tough environment. But universities are built to focus on their primary revenue streams, such as tuition from international students. From an organizational perspective, research is a thing they do to keep their metrics up, to keep that revenue stream coming in from students. As a result, it's not surprising that they’re not really designed for research: as most academics know, getting anything non-standard done within a university bureaucracy is near-impossible.

What is surprising is the degree of Stockholm syndrome many academics feel about their employers: they complain about university and hospital processes, but at the same time often can’t really seem to imagine any other way of working. But there’s no law of the universe that says that research needs to be done in organizations that aren’t built for it.

So I think one way we can break out of this is focusing more funding and energy on organizations that are built from the ground up to facilitate the research process.

Do you mean all of the new organizations like the Arc Institute, Arcadia Science, Focused Research Organizations (FROs), etc.?

Right, these are newer types of organizations that are designed from first principles to tackle a specific type of research problem - in particular, the types of problems that are poorly suited both to traditional academia and to for-profit industry. But there are many other types of organization that are also built with research as their sole driver.

For instance, I hugely enjoyed being at the Broad Institute of MIT and Harvard from 2012-2019, because it really exemplified the idea of designing an organization to support and drive a specific kind of large-scale, transformative research program.

If I talked to the Broad legal department about a new project that involved some uncertainty or risk, their first instinct was to help me rather than just to protect the organization. The driving goal was to get research over the line. That’s something that’s much less apparent in most research organizations, particularly large organizations like universities.

As a result, being at the Broad meant being able to do things that would be incredibly difficult or impossible at other organizations. For instance, my team was able to assemble the largest collection of human genetic variation in the world - the Genome Aggregation Database or gnomAD - because we had incredible technical, financial, and ethics support from many teams across the institute.

The lessons from working at the Broad were a big driver for me in the type of organization I wanted to work with as I was considering moving back to Australia. The Centre ended up being hosted by two amazing non-profit medical research institutes because they had the leadership and flexibility to be able to do something very non-traditional.

Focused research organizations are one end of the distribution, not just built to do research, but for a particular goal/end. But being “built for research” is the key factor.

If you had no constraints in terms of funding or the need to publish, is there anything that would be different about your research?

This is a useful thought experiment!

Firstly, I’ll note that grant-writing isn’t all bad - there is definitely some benefit to the regular process of writing funding proposals, in that it forces us to crystallize our ideas before we’ve done the experiments. That said, it means all of us skew our research direction in ways that lead to funding, and in particular the traditional funding model pushes us all towards safer, less ambitious, shorter-term, and less risky programs.

Having more funding stability provides an opportunity to do things differently. Since I moved back to Australia in 2020 I’ve been incredibly fortunate to have two institutes that have provided a substantial 6-year funding commitment that allowed me to build a large team, create infrastructure for long-term projects, and do things like large-scale community engagement that would have been incredibly challenging with traditional grant funding. This has allowed us to do things that I think will be much more impactful in the long term than what we could ever have done with grant funding alone.

More generally, I think models that provide stable funding to individuals and organizations and allow them to take risks, to build long-term projects, and to quickly pivot to chase the science, are absolutely critical if we want to ensure that we can do the most transformative research. I’d love to see more of this from government funders, but in the interim I’ve been delighted to see many philanthropists and foundations (e.g. Snow Medical in Australia) adopt this philosophy at scale.

What then do you think about the role of soft money, and how that affects the incentives of researchers?

I was on 100% soft money for the whole of my time in the US, and it was pretty terrifying for the first couple of years. Even after I reached the point that I could comfortably keep the group operating with external grants, there was always that point in the back of my head: everyone could lose their salary if I didn’t continue to raise funds, and as the group grew, that pressure magnified.

That said, this pressure isn’t unique to academia: it’s not that different from how small businesses or startups operate. And there is some truth to the idea that the existential need to raise funds concentrates the mind wonderfully on doing the type of research that is appealing to funders, which is obviously good from an organizational perspective.

I don’t really have a strong feeling about the right approach here. I think some level of salary guarantee is helpful, given the inherent uncertainties in research funding, but at the same time I don’t think an academic job should be seen as a guaranteed salary for life. Running an academic lab, like running a startup or a small business, means taking a risk because you believe in the value of what you’re doing. So I think the right model involves some level of funding stability, but also a really clear set of expectations around the milestones and targets you’re expected to hit in order to continue that funding.

What would you change about the organization of science, and how could funders help in that regard?

It’s always a dangerous game to second-guess funders, because of course they think a great deal about who and how they fund, and I’m conscious that researchers can only see some of the aspects of these decisions!

Nonetheless, I’ll take a shot: if I were a funder, I would be thinking as much about the type of organizational structures around researchers as on the specific projects those researchers are pursuing. In general, I’d think about the ways that I could not only drive impactful research, but also how I could drive innovation and change in the way that research is done.

Over the last few years we have seen an explosion of new organizational models for research - we’ve already discussed Focused Research Organizations, Arcadia, and the Arc Institute, for instance. I’d love to see more funders following Ben Reinhardt’s advice and thinking about how they can foster the development and persistence of organizations trying new ways to tackle research problems.

Many of the biggest problems we as a society are currently facing are poorly suited to the traditional academic model (individual labs, mostly composed of trainees with short-term career milestones focused on publications), but also don’t align with the commercial outcomes necessary to build a VC-backed startup. In addition, the pace of technological change we will all experience over the next decade will be blistering: the single most important factor for organizations will be their ability to very rapidly pivot and adapt. So if I were a funder, I’d be thinking very hard about what types of organizations are most likely to be able to rapidly adapt to environmental change and deliver the impact I’m looking for.

Just for fun, what's a paper in your field in the past 1-5 years that you wish you could have published?

I don’t know if these are papers I wish my own group had published, but there are a ton of papers I’m delighted that someone published!

One of the joys of doing open science and sharing data as quickly and openly as possible is seeing other scientists pick up those data sets and use them in really creative ways. So I’ve been thrilled to see the sheer number and diversity of publications that have used the various gnomAD datasets we’ve released over the years - now over 20,000 papers in total.

It’s hard to pick just one or two from a list like that, but I’ll call out a few of my favourites: Molly Przeworski’s team’s really elegant use of gnomAD to understand the pressure of natural selection on individual genes, for instance, or Debbie Marks and colleagues using gnomAD and many other datasets to build a machine learning predictor to help determine which variants are disease-causing, for instance.