X-ray image of the silica shell of the diatom Actinotychus Senarius obtained at 5000-fold magnification. Fine details in the diatom are visible thanks to a powerful new X-ray lens design that is being developed for eventual application to single-molecule imaging.

Image: Andrew Morgan, Sasa Bajt, Henry Chapman, Christian Hamm.

“We are addressing the challenges associated with instrument fluctuations, electronic damage, natural conformational variability and the effects of confinement…by developing new systems and tools to map the heterogeneity landscape of single molecules.”

Single molecule imaging reveals biologically relevant heterogeneities within proteins and protein complexes. Rather than producing an average or dominant structure of a protein or complex, single molecule techniques can capture the‘heterogeneity landscape’ occupied by a single particle. The details of this landscape convey information about biochemical reactions and the functionality of biomolecules within living systems.



Imaging CoE scientists perform single particle experiments that encompass X-ray free-electron laser single particle imaging (XFEL-SPI), cryogenic electron microscopy (cryo-EM), micro electron diffraction (micro-ED) and fluorescence imaging. Each of these techniques faces different and significant challenges in order to achieve single molecule imaging. It is not enough merely to achieve atomic scale molecular imaging, because this has no biological value if the molecule is not in its native state. The real challenge is to adapt and develop our techniques to better accommodate the intrinsic inhomogeneity of the systems we wish to image.

Imaging CoE researchers are advancing this field by combining XFEL-SPI or cryo-EM studies with complementary experiments using crystallography and molecular fluorescence. Such advances can come only though a collaboration spanning physics, chemistry and biology and involving both experiment and theory to determine the real behaviour of the target molecules under imaging conditions.

Our Centre’s expertise in single-molecule microscopy at UNSW, and a technology developed at La Trobe University to produce nano-droplets on demand, will allow us to devise protocols for the appropriate handling of samples in XFEL experiments, delivering molecules in their native states.
In 2018, we engaged with experimental work performed by Professor Ilme Schlicting, a member of our International Scientific Advisory Committee, on the effects of electronic damage on pump-probe XFEL measurements of molecular dynamics. We developed closer links with the Hamburg-based CFEL group led by Professor Henry Chapman (Imaging CoE PI) and recently recruited Dr Andrew Morgan from Hamburg to join the Melbourne node of the Centre.


  1. Utilise topological data analysis to establish the “shape” of large, noisy and incomplete data sets in cryo-EM, XFEL-SPI,
    including the effects of damage on the efficiency of the imaging systems.
  2. Explore electronic damage processes in XFEL imaging techniques.
  3. Develop new sample delivery systems to improve the outcomes of XFEL-SPI experiments.
  4. Explore micro electron diffraction techniques using the Imaging CoE’s expertise across both theoretical analysis
    and experimental studies.


Disordered crystals merge crystallography and single molecule imaging

X-ray crystallography has been the most successful and widely used technique for imaging protein molecules, the building blocks of life. It is well known, however, that crystallography has long suffered from a basic problem: that the information provided by crystal diffraction is insufficient to determine the protein structure. To get around this problem, scientists must rely on prior assumptions about the structure of the protein
or conduct further experiments to compensate for the
missing information.

Another well-known problem in crystallography is that some crystals are observed to diffract “poorly”. In other words, the bright Bragg spots that are emitted, when diffracting X-rays through a crystal, are only observed at shallow angles with respect to the incoming beam and this limits the resolution of the final image. Professor Henry Chapman (CFEL Hamburg, Imaging CoE PI) realised that such “poorly” diffracting crystals may actually provide more information, and at higher resolution, than their perfect counterparts due to the presence of additional diffuse diffraction.
Dr Andrew Morgan joined the Imaging CoE in May 2018 from Henry Chapman’s group in Germany. He has shown that such poorly diffracting crystals can provide enough information to completely determine the protein structure.
This means that many proteins can be imaged without bias and with finer detail than was previously possible at crystallographic beamlines.

A fascinating unsolved problem in science is the mystery of how the Photosystem II protein molecule can so efficiently split water in plants. Being a membrane protein, it is hard to crystallise, and researches often find that it diffracts too poorly to achieve the desired resolution. Dr Andrew Morgan and Imaging CoE physicists are collaborating with groups led by Professor Henry Chapman and Professor Petra Fromme to investigate new and existing Photosystem II datasets to see if this new approach to crystallography can unlock this outstanding mystery.

AB initio phasing of the diffraction of crystals with translational disorder VOLUME: 75 PAGES, Article number: 1 (2019). Andrew J. Morgan, et al, Acta Cryst

A disordered crystal of ducks (top) and the accompanying diffraction pattern (bottom) that exhibits features of both crystallography (Bragg spots) and single particle imaging (continuous diffraction).
Andrew J. Morgan


Searching for order in heterogeneity using mathematical topology

A biomolecule in vivo exists in a wide range of conformations. We encounter heterogeneity in many forms; in the disorder of crystals, in conformal distributions in cryo-EM and in the effects of molecular confinement in XFEL sample delivery by aerosols. We can reduce damage to the sample by rapidly freezing the samples and using weak electron probes, or we can attempt to outrun the damage by measuring structures on femtosecond timescales using brute force XFEL pulses.

Nevertheless, the signatures of high-resolution biomolecular structure are accompanied by high levels of noise in any imaging system. These are the principal challenges of Theme 2; the example below is just one example of our progress in 2018, which also includes the link between noise and resolution within a quantum mechanical formalism of imaging and the development of a hybrid scheme that merges plasma physics and molecular dynamics in describing XFEL-molecule interactions.

All forms of advanced molecular imaging are characterised by a shift to the methods driven by huge datasets. Topological data analysis (TDA) tools have been developed in the last couple of decades to analyse the multidimensional features of such datasets. These tools classify topological spaces using topological invariants such as homotopy, homology, and Euler characteristic. For example, homology can detect topological features such as connected components, holes and voids and record the evolution of these features as the parameters change. Such a feature is extremely useful to classify the data.

We have adapted these mathematical tools with machine-learning techniques to study the conformational space of large bio-molecules and capture their dynamical properties in a compact topological representation. Incorporation of this scheme in imaging algorithms connects heterogeneous structural datasets from distinct conformers continuously rather than classifying them as distinct instances of structure.

Persistent structural features of c1 complement protein (top) in molecular dynamics simulations represented as a vietoris-rips complex (bottom) of topological analysis. The analysis identifies collective motion of subunits within the atomic-scale simulation of the entire protein at varying length scales in an aqueous environment.