Welcome to oxDNA.org!

OxDNA and oxRNA are coarse-grained models specifically developed to simulate nucleic acid nanotechnology. They are parametrized to reproduce basic structural, mechanical, and thermodynamic properties of DNA and RNA. OxDNA is also the name of a software tool implements, among others, the oxDNA and oxRNA models. To learn more about the models, please look at Citations or explore the webpage dna.physics.ox.ac.uk

To start an oxDNA simulation, you need three files: topology (.top) and configuration (.conf or .dat) that describe the structure, and input file that describes the parameters of the simulation (such as temperature, duration of simulation, salt concentration). The format of these files is described in detail here. The server generates automatically the input file for you, but you need to provide the .top and .dat file for the simulation server that describe the configuration you want to simulate.

The TacoxDNA webserver provides a variety of conversion tools from popular DNA nanotechnology design platforms to the oxDNA format. Some tools, such as vHelix, Adenita, and MagicDNA have built in oxDNA outputs. Designs from CaDNAno often need to be relaxed using rigid-body dynamics prior to simulation, this can be done in oxView (hit the "try it!" link in the description).

These simulations are run on NVIDIA 2080 RTX GPUs, which easily handle origami-sized (10-20k bases) designs. Designs with more than 50k bases may not finish in the time allotted to each job. It also makes sense to only use GPU cards for simulations of structures of sizes at least 300 nucleotides, when the GPU provides significant speed-up over CPUs For smaller systems, we recommend submitting them to CPU. They will run just as fast and will leave GPUs available for larger simulations.

For an origami-sized structure running with the default option of 1e9 steps, we are looking typically at about 2 days running time. The running time scales almost linearly with the number of nucleotides, but please note that each structure is different. You can opt into receiving emails when jobs complete on the My Account page.

Currently, it is not possible. The model only support simulations of DNA or RNA molecules alone. We have a model in development that allows you to represent also proteins, but it has not been deployed to the server yet.

Due to storage limitations, the simulated trajectory gets automatically deleted after 2 weeks. You will receive a reminder by e-mail 2 days beforehand, but we strongly recommend to download all data once simulation is finished. The start and final configuration file will remain available on the server, as well as results of analysis that you carried out on the server.

Guest jobs are created using a temporary account which will be inaccessible after logging out. The unique URL for guest jobs (something like oxdna.org/job/<32-digit-uuid>) can be shared and accessed by anybody whether or not they have an account. Jobs submitted by verified user accounts are completley private, only the user that created the job can access the job page.

For security reasons, we do not allow users to submit their own input files.

No. This service is meant to provide simulation support for experimental labs and so only provides equilibrium MD simulations and currently does not support advanced simulation techniques.

Coarse-grained simulations such as oxDNA speed up different processes by different amounts, so it is impossible to put an exact time correspondence between simulation and real experiment. For more discussion on this topic, see e.g. this article. On the server, these simulations will be run by default for 1e9 steps with a timestep of 0.001 simulation time units, which by direct unit conversion corresponds to 3.03 μs. However, due to accelerated diffusion constant that we use, it can be argued that it can roughly correspond to timescale on the order of up to 1 ms. Increasing dt in the simulation parameters can allow you to simulate longer times with the same amount of steps, but can also lead to numerical instabilities and errors. We do not recommend using dt larger than 0.003.

By default, we set the simulation to run for 1e9 steps. This is usually sufficient to get enough statistics to explore behavior of a typical DNA origami-sized structure. However, if your system is more complex and can undergo different conformational transitions, it might not be enough. You should visualize the entire trajectory in oxView and plot energy file versus time and see that each relevant state is sampled, and that energy of system does not diverge, but rather keeps fluctuating around mean value. Similar analysis can be done with distance evaluation, checking that distances between parts of structure of your interest have crossed multiple times between proximal and distant states.

Remember that oxDNA is just a model, which makes numerous approximations to simulate efficiently the nanostructures. While in majority systems that we studied previously, we found quantitative or semiquantitative agreement with available experimental data, you should still take results with a grain of salt and ask yourself: is the behavior seen in simulations something that my system should be able to do? Simulations can never replace experiments, but might help you get insight into the function. If you are in doubt, do not hesitate to contact us or check out some of our publications where we study systems similar to yours for further discussion of underlying phenomena.

In the "Advanced Parameters" section on the submission form there's a file upload for external forces. OxDNA implements a variety of different external force potentials which are defined by specifically-formatted text files. A full list can be found in the documentation.

One of the most popular types of forces is a mutual trap, used, for example, to bring two single-stranded segments into proximity to facilitate binding. These can be generated interactivley in oxView by selecting pairs of nucleotides you would like bound, selecting "Create from selection" in the forces dialog from the Dynamics tab and then downloading the file with the "Forces" button in the oxDNA files section of the File tab.

We recommend that you download the trajectory file and visualize it using oxView. Furthermore, we provide multiple automated analysis options, based on the tools that we developed here. Currently implemented analysis allows you to obtain mean structure and standard deviations from the mean structure to visualize its flexibility. You can also check distances between nucleotides, bond occupancy and angles between different stems in the simulation trajectory. For more complex analysis, you can use our python tools to analyze the trajectory that you downloaded (see tutorials and examples here). The trajectory is compressed in zip format and will need to be inflated prior to use (older trajectories were compressed in 7zip format which, while higher compression, had limited support).

Errors are most frequently due to the model encountering unphysical states because of insufficient relaxation. This often happens when you convert from a different format (most often caDNAno) where the 3D positions of helices are not physical and some of the distances between neighboring nucleotides are unphysically large. Structures exported from design tools are rarely in a physically viable state and must be relaxed prior to production simulation. We have tried to provide a reasonable relaxation protocol automatically, but for some structures it is insufficient. Try resubmitting your job and increasing the length of the relaxation steps or decreasing the dt of the MD relaxation.

Check out this book chapter for a comprehensive introduction and take a look at the documentation for detailed information on installation and running simulations. Also check out our Citation section for the articles describing the models in detail.

On "View Jobs" page, you can see the cluster status and how many nodes are currently occupied and the total number of jobs scheduled in the queue. The jobs are scheduled on first come first serve basis, and you might need to wait until other jobs finish before yours gets scheduled. We currently impose a limit of 4 submitted jobs per user and a maximum running time of 4 days per job. These constraints might change in the future and we reserve the right to limit the allowed maximum running time and number of submitted jobs for users that use the server heavily. While we have plans to expand the capacity of the server in the future, it is currently equipped only with 8 GPUs. Hence, we ask users to be considerate and not use it excessively. For complex simulation tasks (such as FFS) and large number of designs to try, we recommend that you either purchase your own GPU server or use your own university HPC resources. Most universities have GPU servers available for their researchers. The evaluation scripts that are used on our server are available here and you can use them to evaluate and visualize your results from your local HPC service provider.