Visualizing Zika: how we modelled the spreading virus
Creating scientifically accurate models of widespread and deadly viruses is not an easy task. It involves a multifunctional team of computational and structural biologists, virologists, 3D-modellers and visualisators, and a lot of research and prototyping. Here is an insight into the process of creating a 3D model of the Zika virion.
Our team works in the field of biomedical visualization since 2007, and during all those years we have dealt with a great number of problems that occur when you want to visualize biological objects in both scientifically accurate and beautiful way. One of our ongoing projects is called Viral Park. It is a collection of models that sum up all available information about the atomic structures of the most widespread and dangerous human viruses. When we decided to complete our Park with the latest threat — Zika — we faced complications. There was a lack of data on this virus structure and inconsistent scientific researches.
Zika is not a new pathogen — it is known since 1947 when it was discovered in a rhesus monkey in Uganda. Five years later cases of the fever were registered in humans in Uganda and Tanzania, subsequently, outbreaks occurred not only in Africa but in Asia as well. But the virus attracted international attention only in spring 2015, when a new outbreak, first registered in Brazil, spilled over the whole Latin America, and travellers carried the disease to US and Europe.
Zika is especially dangerous to pregnant women and kids. The geography of the outbreaks is associated with a high prevalence of microcephaly in newborns and increased number of the Guillain — Barré syndrome cases in the same regions. Most of the mothers whose children died of microcephaly within 20 hours after birth had fever symptoms during the first or the second trimester of pregnancy. On February 1st 2016 the World Health Organization declared Zika fever an international health emergency. More than one and a half million people were infected since the beginning of the current outbreak. There are no vaccines or medication available against Zika at this moment. By the most optimistic prediction the first Zika vaccine could become available no earlier than the end of the 2016.
The lack of data on Zika that postpones vaccine development has complicated modelling for us as well. We had to base our work on the published information about the structure of related viruses of yellow fever, West Nile and Dengue, which belong to the same systematic group as Zika — Flaviviridae.
To learn more about the pathogens of this group, our team analyzed papers from 1952–2016 that accumulated years of work of molecular biologists, tropical medicine experts and doctors. We collected data from the Journal of General Virology, Lancet, Transactions of the Royal Society of Tropical Medicine and Hygiene, Nucleic Acids Research, Nature Structural & Molecular Biology, Emerging Infectious Diseases and other peer-reviewed journals, as well as CDC and WHO reports. Results of different research groups can contradict each other, and in this case you need to consult virologists and other experts.
Visual Science always wants to visualize viruses at a maximum level of detail and show all the elements of the viral particles. However in cases of some viral components there is no scientific data available — only hypotheses and speculations. We usually try to sum up the most plausible ideas and make visualizations according to them. Also to fill the gaps in structural data computational biologists from the scientific modeling department of Visual Science use methods of structural bioinformatics to predict the possible structures of viral proteins and their interactions. These methods are widely used not only in fundamental research but also in drug development and molecular interaction studies.
The main idea of using computational biology methods for predicting structures of proteins or other molecules is modeling based on similar structures that are already solved by various methods like X-ray crystallography or cryo-EM microscopy. Luckily in case of Zika there was a lot of information about proteins of the closely related Dengue virus and other representatives of the Flaviviridae family. Our main template for modeling the surface proteins was the cryo-EM structure of Dengue virus solved by Dr. Xiaokang Zhang and his colleagues from the team, led by Dr. Hong Zhou from the UCLA, Los Angeles. To make models of the Zika core protein dimers we used data about the similar protein from the West Nile virus. We also used molecular modeling and dynamics to visualize a possible structure of the Zika genome. We made predictions about the spatial structures of separate RNA fragments and modeled their possible interactions with the core proteins.
When we got an understanding of Zika protein structures, we were able to move to assembling the whole virion. At this stage, Visual Science’s 3D-modellers entered the process. They worked closely with the scientific department and adapted some of the molecular structures received by computational biologists to create the 3D-prototype of the virus. Modellers’ main goal was to optimize the surface of all virion parts, insert them in proper places inside the 3D-environment and eliminate intersections.
While prototyping, sometimes our biologists and modelers discover inconsistencies in the published scientific papers. For example, studies that use molecular methodology can sometimes include not very accurate estimations of the numbers of some proteins in the virion. When modelers start to fill the space inside the prototype they find that the particular number of proteins is unable to fit in. That means the model needs adjustment.
When we completed all the adjustments and got a scientifically accurate prototype of the Zika virion, we needed to show it from the most impressive angles and make it entertaining. Our visualization team did this final work. It not only further detailed the model, but color-coded all parts of the structure according to the scientific legend and slit the virion to present its insides.
Visual Science utilizes a number of formats to present its projects. We often use illustrations and video animation, all in high resolution. Another option is to make an interactive 3D-model which allows to turn the object in all directions, select its elements and see annotation to them. That works great for educational and museum exhibition purposes.
We incorporated all mentioned formats in our presentation. This is Visual Science’s animation of the Zika viral particle:
It reflects the present day views on the viral architecture and includes 360 separate surface protein structures, model of the lipid envelope and possible organization of genetic material in complex with capsid proteins. As was mentioned above, we try to use interactive formats such as apps and augmented reality to adapt complex models for educational purposes. For Zika, we chose to make a web app, where the user can explore parts of the virion in detail and use annotations for navigation.
After our model was published the first cryo-EM structure of Zika virus appeared in the Science Magazine, and we got a chance to compare our predictions with the experiment data. Although we didn’t manage to predict the structure of several sites and loops of the Zika surface proteins, the overall model reflects the experimental structure quite accurately. We think that it could even be used in computational design of Zika test-systems or for the analysis of the structural differences between flaviviruses.
Los Angeles Times made a detailed report on our work. New Scientist called Visual Science’s model a “monstrous beauty” and featured it in one of its issues. Other media outlets appreciated our work too — we found it everywhere across the globe: from Forbes Brasil to Al Jazeera and Russian TASS news agency.