Research @ CBIS

The Art and Science of Biomolecular Imaging

Titan Krios, CBIS CryoEM Lab, NUS

BioImaging is much more than simply taking interesting pictures. In fact, image acquisition is but one stage (and not the first) in a logical, linear process. We define the art and science of biomolecular imaging as a four-stage strategy: manipulate, measure, mine and model.

Each stage of the strategy presents challenges and opportunities, requires different tools and techniques, and builds on the previous stage. With the resources of the Centre, and the technical expertise of our staff, we are uniquely qualified to drive research through all stages, fulfilling project objectives and uncovering the previously unseen.


In this stage the biological question is asked, prompting experiments in which molecules, cells, or tissues are modified in some defined manner (such as genetic mutation, application of a drug, or physical interactions with molecules or cells).

interior-vert-slide-lensAll investigations in biology – including imaging – begin with an experiment that modifies the function of a molecule, cell, or tissue. Usually, a foreign gene is introduced into the cell or the function of a resident gene is knocked out. Alternatively, the actions of proteins in cells are modulated by chemicals or the physical environment of the cells.

How the samples to be imaged are manipulated depends largely on the molecules and cells being studied, the probes for tagging molecules, the microscopy method used, and the goal of the research. Cell and tissue samples may be chemically “fixed” or imaged “live” and in environmentally controlled conditions, based on the nature of a particular investigation.

Examples of Samples

In cryoelectron microscopy, protein samples and entire cells are frozen in order to immobilize and visualize systems that are in dynamic motion at more normal temperatures. In light microscopy, individual proteins may be tagged with a fluorescent tag to make them stand out. Samples may also be imaged in microfabricated devices to ensure controlled and reproducible 3D environments, and subjected to various drugs or changes in environmental conditions.


This is the imaging stage, which may involve a range of microscopy techniques.

At the BioImaging Centre, image acquisition can be achieved through a number of microscopy methods including (from lower to higher resolution) nuclear magnetic resonance (NMR) imaging; high-resolution, high-content, high-throughput light microscopy; high-resolution cryoelectron microscopy; and time-resolved atomic force microscopy (AFM). With these tools, we can capture images of cellular and molecular structures down to the nanometer level.

interior-vert-gear-blueFilterSharpening the image

Often, processing is needed to sharpen the acquired image by removing artifacts from the microscope. In confocal microscopes, a pinhole removes out-of-focus photons, while an energy filter in an electron microscope eliminates out-of-focus electrons. One common computational method of processing involves deconvolution, which increases the resolution of light microscopy data and removes the noise and blur that often accompany high rates of data acquisition.


Here, the data is processed, quantified, and analyzed to produce quantitative results.

interior-horz-gear-greenOnce an image has been acquired, it must be analyzed to identify features of interest, calculate 3D structures, track objects with time, and correlate objects in time and space. In the “mine” stage, information about molecules and cells obtained from the imaging stage is combined with biochemical, genetic, genomic, and proteomic information. We look for structural changes in proteins, organelles, cells, and tissues – and changes over time, where possible.

From microscopic cells, enormous data

Biological imaging is computationally intensive work, requiring the ability to accumulate, store, process, analyze, and visualize terabyte-level data sets. That’s why we assembled a high-bandwidth/high-capacity BioImaging Network specially suited to the rigorous demands of image processing and analysis. This network consists of acquisition instruments and high-density computer processors connected by fiber channel to several terabytes of high-performance data storage. We are able to move data at speeds of several gigabytes per second for processing, analysis, and visualization – locally or via remote access.


Finally, the results are represented through computational, mechanical, chemical, and other models.

interior-vert-man-microscopeAfter image data is processed and analyzed, the next challenge involves developing algorithms to model the data and assemble the knowledge gained from the imaging initiative. The quality of the original data determines the accuracy and reproducibility of the resultant model, and a mistake or oversight at any stage will corrupt the ultimate findings. That’s why the BioImaging Centre is fostering the development of expertise in all stages of the process.

An end, and a new beginning

While imaging can be thought of as a linear process, it is also is a self-perpetuating cycle. Feedback gained through the modeling and visualization of images can influence how new samples are manipulated and new images are acquired. By using those results to define new experiments, we seek continual improvement to our strategy, and the most accurate, reliable conclusions possible.