Deep inside molecules—digital twins at the nanoscale

The results provide guidelines for the design of digital twins and triplets equipped with VR. Globally, molecular simulations provide a mature foundation for the creation of digital twin models. Transferring this experience to the virtual dimension is very easy because of the inherent 3D focus of molecular objects. This study present UnityMol as an existing software tool for which VR porting has been successful and experiments can now be routinely performed in VR with enhanced digital twin models. Several use cases were presented to demonstrate this, starting with the creation of molecular models and integrative models in particular. An intuitive VR tool could help this thriving field by bringing molecular simulations to scientists from other fields who are not experts in these methods. This also applies to the I3 rational design of molecular devices, such as the artificial water-filtering channels presented in this work. Another direction has been proposed by physically representing twins with tangible 3D-printed devices to create real-world artifacts. This approach is still in its early stages and we have presented important concepts. In summary, static exploration and molecular assembly have been well developed, and only the usability and number of functions need to be improved. Interactive simulations are also quite advanced in their implementation but still have fundamental problems, such as an inherent barrier on the time scale, which we will discuss below. For molecular triplets, we are still in the early stages. For example, some physical limitations on the fabrication and construction of 3D-printed models are yet to be resolved. The difficulty in using batteries to power the sensors is a particular challenge. The authors are currently considering several possible improvements to the design, including transmitting power remotely, as in wireless inductive charging, or using a powered base station to position molecules in the workspace.

The time scale is an important issue in molecular modeling. Not all simulations can be performed in an 336 Virtual Reality & Intelligent Hardware August(2022) Vol. 4 No. 4 interactive environment because it would take too long to perform the necessary calculations. The interactive VR experiment with a digital twin is limited by the time a human operator is willing (and able) to control the simulation. This is typically from a few minutes up to half an hour. For example, a molecular dynamics simulation computes only a few nanoseconds or perhaps a microsecond of the dynamics of a molecule. Therefore, the magnitude of the forces that the user applies to the simulation to produce a visible effect is usually too large compared with the forces that occur in real biological systems without causing damage. Even when the simulation is run with massive computational resources, this limitation is severe. A classic solution is to create a simplified system in which the degrees of freedom that are deemed unnecessary are removed. This model simplification–also called coarse-graining–can be achieved by using implicit solvents and membranes or by coarse-graining the representation of atoms by grouping them into larger grains. Such simplification saves significant computational resources and improves operational efficiency. Natural constraints arise from the fact that simulation results must be accurate enough for the relevant application. This requires the identification of degrees of freedom that do not affect the desired property and can be neglected. Another possibility is to perform interactive manipulation with unrealistic forces and then couple it with a longer noninteractive follow-up simulation that allows the use of gentler forces. Such a procedure can be viewed as informed planning and optimization of the simulation process. The author experimented with this in the context of the outer membrane protein FepA, where they sought a passage to drive a transport molecule through a narrow inner channel. Such a passage can be sketched in an interactive simulation, its path recorded, and then replayed in a classical controlled simulation to guide the molecule more smoothly through the system.

An important issue in these experiments is the contribution of the human operator. It has been shown that interactive simulations, such as those discussed for RNA folding in the classroom, also have great potential for research. This project can be viewed as an online 3D puzzle in which players are asked to shake and wiggle the 3D structure of proteins to find the most stable conformations. Such a collaborative puzzle can use the power of the crowd to solve complicated scientific problems; hence, the approach is called “crowdsourcing”. The project has produced remarkable results from a biological perspective, but has also been useful for the collaborative development of new algorithms. The problem addressed by FoldIt is similarly complex, that is, NP complete, as in the drug design example this study presented. Similar to our classroom example, the analysis of the tasks performed by the participants showed that the players took a different path from that taken by the computer. The analysis showed that there was no equivalent algorithm for player behavior. This is one of the main reasons for implementing I3 rational design approaches that are intuitive, interactive, and immersive, allowing the use of these “human algorithms”. Generally, interactive simulations and citizen science are areas that are rapidly developing and where the number of projects shows significant growth, including in the field of biochemistry. Digital twins can eventually be placed in the hands of fellow citizens to provide deeper insights into many processes, whether in nature or in industry.

The concept of using augmented and virtual elements to connect the physical “real” world with the invisible and opaque nanoworld promises highly intuitive interfaces. This is achieved by creating digital, physical, and tangible triplets. This approach is based on intelligent hardware that creates a conduit between the digital model and real tactile manipulation. It borrows heavily from the concept of data physicalization and extends it to perform live simulations. The current prototype establishes a one-way connection between the physical object and digital domain. In the future, it would be useful to establish feedback communication that connects simulation data to the physical model, possibly in the form of vibrations or other tactile feedback.

Two main themes have emerged regarding future directions and development characteristics. First, the collaborative aspects should be refined so that multiuser scenarios can be implemented in which complex tasks can be completed using digital twins. One particular application area mentioned earlier is integrative mod-Marc BAADEN. Deep inside molecules — digital twins at the nanoscale 337 eling, as a number of conditions must frequently be met simultaneously to create a valid model, which may require simultaneous, collaborative control by multiple users. Second, technologies used for interaction and feedback, such as haptics in immersive environments, should be developed beyond the currently supported device capabilities. They do not presently allow the user to “feel” the properties of the molecular materials of the digital twin with sufficient depth, although the mechanical properties are of great importance.

In conclusion, the molecular domain offers tremendous opportunities for digital twins in VR environments if most of the technical and technological hurdles could be overcome. Once the onboarding of VR is simplified and the technology is widely adopted, this approach should find a much wider range of applications.