This overlay of mass spectrometry images shows the spatial distribution of three different kind of lipids across a whole mouse cross-section. Lipids act as the structural components of cell membranes and are responsible for energy storage, among other things. | Photo courtesy of Wolfgang Reindl (Berkeley Lab).
With the school year just getting underway, teachers are figuring out the learning styles of their students -- whether they are obstinate learners (which parents might hear more about at the next parent-teacher conference), linguistic learners who prefer words, or visual learners who prefer pictures and images.
Visuals and images are important in the sciences, too. They’re especially useful when the original data set looks a bit like the squiggles made from trying to get ink to come out of a pen. And that’s what makes OpenMSI, a web-based platform recently established by the Biosciences and Computational Research areas at Lawrence Berkeley National Laboratory (Berkeley Lab), so important.
The “MSI” part of “OpenMSI” stands for mass spectrometry imaging. Essentially, mass spectrometry is a tool that scientists use to figure out what building blocks -- atoms and molecules -- something is made of, how those building blocks stack together, and how much of each building block is there. For instance, a researcher might put a droplet through a mass spectrometer and find out that it contains a molecule with the chemical formula C8H10N4O2, and realize that it’s caffeine from coffee spilled on the sample.
MSI takes mass spectrometry much further. It transforms the squiggled peaks of those data sets into full-color maps of a sample, with each type of atom or molecule represented by a different shade. Using MSI, scientists can determine the makeup of those molecules, and then visualize how those molecules are spatially distributed. That makes it a powerful tool for everything from developing more effective medicines to creating better biofuels.
However, MSI’s data sets are massive, typically ranging from 10-50 gigabytes, the equivalent of some 20,000 digital photos. That makes them clunky to manipulate and challenging to share, both significant obstacles given the collaborative nature of science.
And that’s where Berkeley Lab’s OpenMSI comes in. OpenMSI, developed by lead researchers Oliver Ruebel and Ben Bowen, is a gateway that uses the National Energy Research Scientific Computing Center’s (NERSC) supercomputing resources to process, analyze, store and serve up MSI data sets to researchers. OpenMSI is web-based, so scientists can simply use their web browsers to study different samples in real-time. The gateway allows researchers to retrieve massive MSI data sets in the blink of an eye (less than .3 seconds, to be precise), and its user-friendly interface means that even scientists without programming skills can visualize, analyze and share their data.
OpenMSI already has several detailed data sets up, and more are expected to be added in the coming months. Moreover, there’s plenty of power at NERSC on which to draw, including Hopper, a supercomputer capable of performing more than a quadrillion operations every second. That means that the possibilities of OpenMSI are just beginning to be explored.
Learning styles may be different, but what really matters … is the grade. Actually, as teachers will (eventually) admit, what truly matters is not the letter, but rather the lesson; the learning that lives across a lifetime. That’s true of efforts at Berkeley Lab, and across the Energy Department, as well. OpenMSI is a gateway, and there’s every hope that researchers will use it to provide life-long-lasting benefits.