Shaping the future of science
Open source data tools, pioneered by Professor Jason Swedlow, are revolutionising research. [Image credits: University of Dundee]
To say that Professor Jason Swedlow has had a profound impact on science would be an understatement.
At a time when biologists are drowning under data, he has delivered software and protocols to store, share and analyse the intricate, digital microscope images that are being used to understand the brain, fight viruses and beat cancer.
And importantly, these tools are open and free.
“People will tell you that in science you have to compete and you have to win, and for sure, there is an element of that,” Swedlow tells Microscopy and Analysis. “But I have always felt very strongly that the best science is done through collaboration and connection, so putting together these open-source developments and tools is how I like to work.”
From word go, Swedlow was drawn to practical science. At school, he wanted to understand how parts fitted together, so chemistry became an enduring passion.
In 1982, he graduated with a degree in Chemistry from Brandeis University in Waltham, Massachusetts.
As he says: “Scientific education can be pretty painful as you sit in lecture after lecture listening to a tonne of information that you need to learn to go forward.”
“But at Brandeis we had this opportunity to get into a research laboratory very early on to see exactly how [science] happens, which was just fabulous,” he adds. “So this is where it all came together for me.”
From Brandeis, Swedlow's scientific career took a detour as he juggled amateur bicycle racing with studying circadian rhythms in mammals at Massachusetts Hospital.
Drawn to the California sunshine, he soon crossed the US to join a racing team as well as the Laboratory of Harold Papkoff at the University of California, San Francisco.
At the time, Papkoff, a comparative endocrinologist and biochemist, was preparing, as Swedlow puts it, 'the most potent and pure gonadotropins and prolactins'.
The young researcher learned a lot about purifying proteins, and importantly, the work paid for his tyres and food.
Still, come 1985, Swedlow's attentions turned purely to science. As he says: “Cycling was so intense and I was competing against sponsored racers that didn't have to work - I realised I really needed to do something else.”
3D Structured Illumination Microscopy with OMX: a 3D representation of nuclei in a late stage of nuclear division, from different sides. [Dundee University]
At the time, Professors David Agard and John Sedat, from Biochemistry and Biophysics at UCSF, were pushing back the boundaries of structural biology and determining the organisation of DNA and chromosomes within Drosophila nuclei.
What's more, the researchers were using a CCD camera with deconvolution fluorescence microscopy to image these biological structures in 3D.
“I interviewed with Dave and found out that they were doing 3D microscopy; it was just the craziest thing,” he says. “They'd got this CCD camera, which at the time no-one but the US Defense industry had, and hooked it up to a custom Zeiss Axiomat fluorescence microscope and were actually taking 3D images of cells - it was mind-blowing and no-one had done this.”
So, with Agard and Sedat as his supervisors, Swedlow embarked on his PhD, investigating the dynamics of DNA Topoisomerase II enzymes in Drosophila chromosomes.
“The next phase of this research was to take images of live cells which at the time was unbelievable - the microscope developed by Agard and Sedat were one of the few that could routinely perform live imaging of cells and embryos,” highlights Swedlow. “Wide field microscopy combined with deconvolution was relatively fast and light efficient, so we were able to perform several unique live imaging experiments.
“Collaborators came from all over the world to access the technology - it was a great environment to learn about this and biology,” he adds.
By 1994 Swedlow had completed his PhD and started post-doctoral research into chromosome dynamics with Dr Tim Mitchison at UCSF. Mitchison moved his laboratory to Harvard Medical School in 1997, and then a year later, Swedlow left the US for Scotland.
At the time The Wellcome Trust was encouraging up and coming researchers to move to the UK, and according to Swedlow, the University of Dundee was one of the new exciting centres for biological research.
“I didn't ever imagine leaving the US but was very impressed with the people at Dundee, and how they communicated and shared ideas,” he says. “This was very similar to the environment at UCSF but I was also attracted to the European model of shared resources and capabilities.”
So in 1998, Swedlow set up his own laboratory as a Wellcome Trust Career Development Fellow at the University's Wellcome Trust Biocentre. It didn't take long.
“We had these amazing research and support facilities so within a few weeks I was doing experiments,” he says.
At the time, a big question on biochemists' minds was how one of the five key proteins in the eukaryotic cell, histone H3, was responsible for the structure of the DNA and protein complex, chromatin.
As part of this, many researchers including Swedlow, wanted to know exactly how this histone phosphorylated during cell mitosis.
3D representation of a living cell during the process of mitosis. [Dundee University]
“So many research groups were working on this,” says Swedlow. “But Dundee had a wealth of facilities, really powerful signalling capabilities and a lot of expertise in tracking phosphorylation sites, so Dundee was a great place to work this out.”
Success ensued. Swedlow and colleagues soon identified an Aurora kinase and protein phosphatase 1 as key players in histone H3 phosphorylation.
They went on to identify new Aurora B substrates and also identified a novel protein, Bod1, that played a critical role in regulating the phosphorylation of mitotic proteins.
Then, in 2013, they discovered that Bod1 modulated the activity of a different mitoic phosphatase, protein phosphatase 2A, and later showed that Bod1 had functions outside of cell division and was required in neurons as well.
The image data resource collects biological imaging data around the world: Data for “RNAi screens for Rho GTPase regulators of cell shape and YAP/TAZ localisation in triple negative breast cancer” by Pascual-Vargas et al.
These breakthroughs have since had a huge impact on cancer research, shedding light on how cancer cells divide and also opening the door to the development of new cancer therapies.
And today, Swedlow's laboratory continues to investigate how Bod1 work in neurons, using a variety of imaging and other approaches. They are also exploring the function of other protein modifications in the cell cycle, including proline hydroxylation.
Dedicated to deconvolution
Even from his early studies of mitosis, Swedlow had relied on deconvolution microscopy, initially imaging cells using a DeltaVision instrument from Applied Precision Inc. and more recently using an OMX microscope, originally commercialised by the same company (see 'Brave new platform').
“[The DeltaVision] was our workhorse for a whole bunch of reasons,” highlights Swedlow. “It could achieve what was good resolution at the time and was very sensitive so we could do a lot of live cell imaging.”
Since, the late 1980s, laser confocal microscopy had also been under rapid development.
Given this, Swedlow, along with colleagues, John Murray and Jennifer Waters, devoted more than a decade to looking at the performance of the two technologies, comparing photon efficiency collection, image contrast and signal-to-noise ratio at equivalent illuminations.
"We are working on our dream of making imaging data as valuable as genomics." Jason Swedlow
Along the way they showed that the performance of any imaging system was a combination of the specific imaging modality, such as wide-field spinning disc, or laser scanning confocal microscopies, along with the implementation choices made by the manufacturer.
But as Swedlow says: “What really came out of this is that no single modality solves all problems in biology and you really need to think about what you want to measure from your biological specimen to decide which modality to use.”
OME develops opensource software and data format standards for the storage and manipulation of biological microscopy data. [OME]
Still, given their successes with deconvolution microscopy, Swedlow and colleagues wanted to use the method to delve deeper into cell biology, and come the early 2000s research into the mitotic spindle was underway.
This macromolecular machine segregates chromosomes to two daughter cells during mitosis, and as this is a dynamic process, live cell imaging became increasingly important for probing the molecular mechanisms that assemble, regulate and drive the spindle machine.
At around the same time, Swedlow was collaborating with Dundee colleague Angus Lamond and using live cell imaging to measure the dynamics of structures within the cell nucleus.
Like many in the field, the researchers were awash with large, unwieldy datasets that had to be processed and analysed.
As a result, Swedlow started to collaborate with Peter Sorger, then Professor of Biology and Biological Engineering at MIT and Ilya Goldberg, at the time a postdoc in Sorger’s lab, to develop an image-management system to handle the vast swathes of data pouring from cell studies. Open Microscopy Environment (OME) was born.
OMERO.insight from OME provides tools for viewing and managing data in an OMERO Server. In this example, users can adjust image rendering without altering raw data. [OME]
“In our research we were thinking about larger image-based assays, which led us to this question of how do you handle and analyse all the data,” says Swedlow. “So we started to build open source software to [automatically] analyse, model and mine large image sets.”
“We showed it worked with our live cell imaging data, and then published a white paper in Science [in 2003] describing our ideas, and this became the snowball rolling down the hill,” he adds.
With the concept demonstrated, Swedlow and colleagues from Dundee swiftly started to develop one of OME's key tools, OMERO, an open source data management platform for accessing, sharing and analysing complex multi-dimensional image data.
Around the same time, Kevin Eliceiri and Curtis Rueden at the University of Wisconsin, Madison, joined OME, building and releasing 'Bio-Formats', a plug-in library for reading and writing microscopy file formats.
OME also released OME-TIFF, an open file format for multi-dimensional imaging. These open source projects have all been planned and developed in full view of the wider community, with the tools now being used by thousands of researchers worldwide.
Along the way, Swedlow launched Glencoe Software to provide commercial licenses and support to companies, and also to customise OME's software tools.
As he puts it: “Many biopharma and technology companies wanted a commercial licence, rather than an open-source licence... so we formed Glencoe to get the technology as widely tested and distributed as possible.”
“One of my colleagues always jokes that Glencoe isn't about yachts in Monaco, it really is about finding ways to exercise the technology,” he adds.
Glencoe Software provides commercial licenses to companies – in one example, FEI incorporated Bio-Formats Library into its 3D platform, AMIRA. [Glencoe]
With Glencoe Software formed, Applied Precision, PerkinElmer and Rockefeller University Press were quick to take on OME's tools.
Indeed, in 2008 Rockefeller University Press launched 'JCB Data Viewer', which based on Bio-formats and OMERO, allowed users to browse image data linked to Journal of Cell Biology articles (see image of 'Whole zebrafish embryo'].
“With JCB Data Viewer, researchers were publishing hundreds of gigabytes of data online with a paper for the first time,” says Swedlow. “At the same time PerkinElmer picked up the technology and turned it into the commercial product Columbus; this was so incredibly important to us.”
Whole Zebrafish embryo: In 2012, an updated version of JCB DataViewer was launched alongside a 281 Gigapixel image of a zebrafish embryo. The electron microscope image comprised 26,000 tiled images and swiftly became viral, sustaining around 300 hits/sec within hours of being published. Here, overlapping images are shown at progressively higher magnification to show subcellular features within single-cell, tissue, and organismal contexts. Red boxes indicate the region selected for each successive magnification step (indicated as percent magnification). The maximum original image resolution of 1.6 nm per pixel is shown as 100%. In 2018, this image was moved to the IDR and the rest of the JCB DataViewer datasets were moved to the BioStudies database at EMBL-EBI.
Crucially, in 2015, Swedlow and colleagues at the European Bioinformatics Institute (EMBL-EBI) won a hefty £1.79 million from the UK's Biotechnology and Biological Sciences Research Council to use OME tools to build a public image data repository.
The end result has been the Image Data Resource (IDR), a vast, public database that collects and integrates biological imaging data from scientists worldwide, was unveiled.
Importantly, IDR has provided a journal-independent data publishing platform that routinely publishes Terabyte-scale images that are linked to published papers.
These can be accessed and re-used by the global scientific community. IDR is now working with Euro-BioImaging, and aims to be part of this developing European research infrastructure’s programme of Data Resources.
“I firmly believe that open, accessible technology development is a key part of the scientific enterprise," highlights Swedlow.
"New technology moves science forward, and developing technology in the open allows it to be tested, validated and advanced alongside real scientific applications,” he adds. “With projects such as Euro-Bioimaging, we are building communities that are sharing expertise from around the world; performing science in this way has been one of the most rewarding parts of my career.”
Swedlow's words on the Image Data Resource echo the very reason that the OME originally came into being; to develop open-source tools to store and share images. And clearly this goal is being achieved on an ever-larger scale.
But from outset, Swedlow has also likened sharing image data to sharing genome data; is he getting closer? It would seem so.
Just last year, Swedlow and OME colleagues from Dundee University joined the global effort to develop a Human Cell Atlas and map every single cell in the human body in a bid to fight disease.
“Genomics is so important to biologists and life scientists,” he says. “We now know that public image data resources can be built and are valuable, and we are working on our dream of making imaging data as valuable as genomics.”
Brave New Platform
The world’s first widefield deconvolution microscopy, OM0 was built around a Zeiss Axiomat by Yasushi Hiraoka, John Sedat and Dave Agard from University California, San Francisco, and colleagues.
They used the instrument to obtain their first 3D digital fluorescence images of cellular structures in 1980s.
The next version built by Sedat, Agard and colleagues, OM1, was built around an Olympus inverted microscope and designed for routine use, and was also the forerunner of the DeltaVision microscope, built and sold by Applied Precision, Inc.
“It was definitely a workhorse, especially for live imaging”, says Swedlow.
The OMX was then built by John Sedat to support Mats Gustafsson’s 3D Structure Illumination Microscopy.
“This was a really beautiful fluorescence microscope with a really important technology bolted onto it,” highlights Swedlow.
OMX image: Retinal pigment epithelial cell in mitosis [Markus Posch, University of Dundee]
This was also commercialised by Applied Precision, and then in late 2008, with support from the Scottish Government, an OMX version 2 was installed in Dundee and made available as a resource for Scottish, UK and European life scientists.
Come 2013, with £1 million funding from the UK research councils, the Dundee OMX was upgraded to a v4 “Blaze” version, which enabled 3D structured illumination microscopy in time-lapse mode.
And since this upgrade the system has continued to be a national resource, pushing back the boundaries of cell biology studies and contributing to major breakthroughs including Swedlow's cell mitosis discoveries.
“A condition of the funds has been that the instrument is open to use for all researchers - this has been brilliant as researchers from all over the world, with different problems and biological systems, have come here to use it,” says Swedlow.