Creating a TI method

To wrap a TI method, you first need to understand how dynwrap wraps a trajectory model

Once you know how to do that, there are three ways to wrap a TI method:

  • Directly through R: Using R packages as dependencies, and using R to read in the data, infer the trajectory and return a trajectory model. This has the least overhead, although you’re restricted to R.
  • Using a wrapper script: A wrapper script in python/R/… that will read the data, infer the trajectory and save the output
  • Using a container: A container in which any dependencies are installed, and which calls a wrapper script internally. This has the largest learning curve, but is the easiest to use on different execution environments and the easiest to distribute to users.