Information for model authors

Despite our efforts to produce reliable virtual experiments with this system, unexpected behaviour may not necessarily reflect a real consequence of the model. Mathematical singularities or other numerical simulation issues may cause the simulated experiment to fail. Sometimes the published representation of the model is in error, or its CellML encoding is. On other occasions the protocol, especially the post-processing section, may need further refinement to account for raw simulation results that fall somewhat outside the expected regime — computing a robust APD that accounts for any shape of AP, particularly pathological cases, is surprisingly complex.

If therefore you find differences in behaviour between the Web Lab and other model implementations, there are several likely avenues for investigation.

1. Check any error messages produced by the virtual experiment run. These will be listed (in brief) in a file called 'errors.txt' as part of the experiment's results, under the 'Experiment information' section. If no graphs are produced by an experiment, this file is displayed by default (if it exists). Often the full textual output from the simulator is necessary to understand fully what caused the error, and this can be found in the 'stdout.txt' file. Unfortunately some errors are rather opaque to new users (although we are working to improve this), so do feel free to contact us if anything is unclear.
2. Check that the metadata annotations on the model are correct. These metadata annotations tell the Web Lab which variables in the model encode particular biophysical quantities of interest. It may be that a variable has not been annotated when it should be, so for instance a concentration clamp might not be applied. It may be that an annotation has been used on a variable which doesn't actually represent the biophysical quantity concerned, and a different variable should be annotated instead.
   If the model is one you have created, the easiest way to check annotations is using the 'edit metadata' tool (the icon). Otherwise you will need to view the model as CellML and look for 'bqbiol:is' content.
   More information about these annotations can be found on the Chaste wiki.
3. Check the CellML representation of the model's mathematics. It may be that the equations or parameters in the CellML file don't match the original model implementation. Since reading CellML directly is rather painful, we advise downloading the CellML file from the Web Lab and loading it in OpenCor, an open source CellML editor and simulator.
4. Contact the Web Lab team. We always invite users who find any examples of behaviour that do not match other model implementations to contact us and we will attempt to determine the cause.