COMPANY CONFIDENTIAL. DO NOT DISCLOSE.
With NanoSIMS technology, scientists reconstruct the sub-cellular
Images generated by NanoSIMS at the UK’s National Physical Laboratory are recreated with data visualization software so that scientists can explore – in detail – analytical regions smaller than a mammalian cell. More efficient analysis helps scientists maximize scientific impact.
Challenge: Analyzing and imaging cells using multi-isotope imaging mass spectrometry (MIMS) with a CAMECA NanoSIMS instrument can generate massive amounts of raw quantitative image data. Data preparation and analysis is therefore highly time-intensive.
Solution: Research scientist Greg McMahon exports data from the NanoSIMS instrument into JMP, where he uses the software’s Graph Builder platform with dynamic linking to perform statistical data analysis and mine image data for areas of interest. With just a handful of simple scripts, McMahon has automated repeat analyses so that he can now go from data to results in just a few clicks.
Results: With its capacity for both handling large amounts of data easily and enabling visual exploration, McMahon says JMP has significantly reduced data preparation and analysis time. More reducible data outputs and efficient workflows are helping NPL and its partners to optimize the utilization of the NanoSIMS instrument and drive maximum scientific impact.
_____________________________________________________________________________________________
At the heart of all scientific research is the ability to accurately and precisely collect measurements. In the UK, the National Physical Laboratory (NPL) is a research leader in measurement science, providing world-leading measurement solutions to both academic and industry partners. The NPL’s metrology experts deploy state-of-the-art technology for wide-ranging scientific research, from evaluating the surface structures of materials for electronic applications to monitoring how targeted therapies reach tumor cells in the body.
Among the NPL’s many research services is SIMS, a technology traditionally used in geochemistry and semiconductor applications to allow scientists to analyze the isotopic composition of surfaces. The technique focuses a beam of primary ions at the surface of a sample, and the impact causes atoms and molecules to be ejected from that sample surface, a small fraction of which are ionized in the process and termed secondary ions. Ultimately, secondary ions are the key to the next analytical step as these charged particles can be directed into a mass spectrometer and sorted by their mass-to-charge ratio.
The mass spectrometer uses a magnet to filter ions by their mass in a similar manner as a prism is able to filter light into the different colors of the rainbow, explains Greg McMahon, PhD, Director of NPL’s NanoSIMS laboratory. “A prism separates light into colors, and the magnet in SIMS separates the particles by their mass and charge,” he says. These secondary ions are characteristic of the region from which they were emitted, thus allowing for the mapping and imaging of surfaces.
Higher-resolution instruments allow imaging of sub-cellular structures
McMahon explains that the next level of SIMS technology – nanoscale secondary ion mass spectrometry (NanoSIMS) – has significantly higher resolution than traditional SIMS and therefore can be applied to answer more challenging problems in single-cell or subcellular biology. Furthermore, it is equipped with seven separate detectors, meaning that multiple isotopes emitted from within the same analytical region can be detected. Scientists can then use information from multiple isotopes to build a map smaller than a mammalian cell. At 35 nanometer side-to-side resolution, scientists can, for example, pinpoint not just which cell is taking up isotopically enriched molecules, but where within the cell those molecules accumulate.
The applications of NanoSIMS technology are vast. Among them are the advancement of pharmaceutical and biotechnology research, which has been a significant area of focus for NPL. Today, much of McMahon’s work involves collaborating with life sciences institutions and companies in the UK to, for example, address problems like drug attrition. “Sometimes millions are spent in R&D for a particular promising candidate compound, only to have it fail at the very end during clinical trials. So if it is to fail, it’s better to fail fast,” McMahon explains. “If we can help pharmaceutical companies decide what’s not going to work [at an earlier stage in the drug development life cycle] and do so quickly, we can save them a ton of money.” Perhaps even more importantly, early evidence that a drug candidate will fail helps scientists redirect their resources and time to other more promising drug candidates.
> Read more about NanoSIMS biological applications in McMahon’s article, “High-Resolution Multi-Isotope Imaging Mass Spectrometry (MIMS) Imaging Applications in Stem Cell Biology,” in Current Protocols
Optimizing the utilization of an extraordinarily expensive scientific instrument
As compelling as NanoSIMS technology may be, instruments like the one at NPL are rare. The technology’s high cost puts it out of reach of many organizations, and as of 2021 there are only about 40 NanoSIMS instruments in operation around the world. To drive maximum scientific impact, scientists working with NanoSIMS are therefore under pressure to shorten experimentation and analysis times.
McMahon explains that at NPL, data collection on the NanoSIMS instrument can in some cases take 18 hours or more for a single run. Traditionally, a standard rule of thumb was that the lab could expect to invest three times as many hours on analysis as on data collection; if one run took 18 hours, data analysis could take more than three workdays. But with an instrument as expensive to run and maintain as the lab’s CAMECA, time lost to analysis means that less time can be devoted to running the instrument in service of science.
Image data, however, poses a number of unique challenges. For one, McMahon explains, because every pixel in an image is meaningful, there are endless opportunities to keep collecting increasingly minute data. “Our regions of interest start to get smaller and smaller and smaller and still carry meaningful data,” he says; that means larger and larger data sets that can be a struggle for software ill-equipped to handle large file sizes.
To optimize the instrument’s utilization, McMahon recognized the potential of a statistical software to streamline both experimentation and analysis that had previously been done with the use of Excel. Even with the ability to code advanced Excel macros, McMahon says a lot of time was lost to time-intensive data preparation steps, data access, converting between formats, repeat analyses and the inevitable troubleshooting that came along with it. Furthermore, he saw the need for better, more interactive visualizations with which to more quickly identify outliers in image data.
A better approach to exploring image data
With its combination of user-friendly visualization features and capabilities for more advanced analysis, JMP was an obvious choice. “When I first started using JMP, it was just, ‘wow, this is fantastic’ – it was easy and well organized and just made a lot of sense,” McMahon explains. The ability of JMP to process very large data sets was an especially significant innovation compared with previous standard approaches. And visual graphics help capture image data in its native format. “That’s where I think JMP really shines – in the visualizations it produces,” he says.
By exporting the X and Y pixel locations and counts from raw NanoSIMS data directly to JMP, McMahon can reconstruct image data in Graph Builder, a platform that also supports data manipulation and analysis. Creating subsets of data that represent a single region in the image is done with one click, and the software’s dynamic linking feature allows users to perform a wide range of analyses, all within the same tool. “That really makes the exploration part so much easier: If you have your image built in Graph Builder, then you've got a distribution of the masses that you're measuring,” says McMahon.
“We can select the outliers [like an isotopically tagged cluster, for example] by just looking at where they're located in the image. It's a very, very quick and efficient way of doing things... JMP is really flexible in terms of data exploration.”
With just a small amount of scripting in JMP, automated analyses get results in ‘just a couple of clicks’
Since the introduction of JMP, McMahon reports that the tool’s Graph Builder platform alone has significantly reduced analysis time, and automation has amplified the benefits. The ability to script and to do it easily makes JMP customizable which means that the data processing aspect can be largely automated and will save time and provide reproducible data outputs. All of this allows scientists to solve problems faster.
“We've got [the analysis] automated to a fair extent now where we just run the scripts most of time. So it's just a couple of clicks and we’ve got the answer right there,” McMahon says. “And the learning curve for scripting in JMP is quite reasonable, in part because JMP has a lot of online resources.”
In addition to doing more with customization, McMahon plans to use those online resources to explore new ways to put JMP to work, be it through design of experiments (DOE) or other modeling applications. For one, McMahon suggests that an instrument operator could use DOE to predict the effect of changing each of the lenses on the NanoSIMS instrument and therefore pick the instrument’s optimal alignment without manual testing. Comparison of the predictive data with actually collected data showed excellent correlation.
“I’ve actually switched some ways I operate things based on the results of design of experiments,” he says. Opportunities to work more efficiently could mean hastening the pace of high-impact science, and that’s what discovery is all about.
