1 of 1

Rule 1

Define context clearly: Develop and document the subject, purpose, and intended use(s) of the model or simulation.

Rule 2

Use appropriate data: Employ relevant and traceable information in the development or operation of a model or simulation.

Rule 3

Evaluate within context: Verification, validation, uncertainty quantification, and sensitivity analysis of the model or simulation with respect to experimental reality and to intended use.

Rule 4

List limitations explicitly: Identify restrictions, constraints, qualifications on the use of the model.

Rule 5

Use version control: Be able to trace the time history of modeling progress including parameter provenance and contributors' efforts.

Rule 6

Document thoroughly: Maintain laboratory notebooks with up-to-date informative records code development, model mark-up, model use and testing, etc. Develop user and developer guides and tutorials.

Rule 7

Disseminate broadly: Publish and make downloadable all components of activities, including simulation software, models, simulation scenarios and results.

Rule 8

Get independent reviews: Have the model reviewed by nonpartisan third-party users and developers.

Rule 9

Test competing implementations: Use different execution strategies, implementations, software packages to check the results.

Rule 10

Conform to standards: Adopt and promote generally applicable and discipline specific operating procedures, guidelines, and regulations accepted as best practices.

Example:

Mechanistic MSM for cortical hyperexcitability

1. context clear?

We were interested in assessing variants of hyperexcitability in an exploratory manner. We therefore did not pre-define a clear clinical context but left ourselves open to exploration of cortical activation for dystonia and seizures.

2. appropriate data?

Data used to develop the model was taken from a large number of sources including different species, different preparations (slice, cell culture, in vivo, ex vivo), different age animals, different states, different conditions. None of the data was taken from the clinical disorders in question (limitations of human experimentation).

3. within a context?

The model lacked the major output for dystonia validation, motor output, so could not be evaluated in proper context. Epilepsy context was somewhat more clear since based on cortical activity level. Beta activation in cortex was used as a dystonia evaluation surrogate.

4. limitations explicit?

Model does not contain representations of spinal cord or limb. Model is lacking many relevant pharmacological parameters, particularly with respect to role of neuromodulators (known unknown), brain states (less known unknown) and metabolic parameters.

5. version control

Mercurial (hg) used for all simulations as well as for paper and figure versioning

6. documented?

Model available with README on ModelDB (189154). Partial parameter provenance and documentation available in manuscript. Full provenance database lacking.

7. disseminated?

Model code on ModelDB available for manuscript reviewers on request. Model freely available after publication. Model and data disseminated via meetings. Manuscript published in open-source journal. Previous models have been used and modified by others.

8. independent reviews?

This model was not reviewed by manuscript reviewers. Model was reviewed for run-capability on multiple platforms by ModelDB curator (Tom Morse) prior to release.

9. competing implementations?

No other implementation attempted. Partially comparable implementations for epilepsy in neocortex do exist (e.g. Traub models, our prior models) and activity patterns could be compared.

10. conformed to standards?

We conformed to standards for ModelDB submission and to best practice standards as taught in NEURON courses. Simulation workflows, data collection, data processing, data reporting conformed to practices generally accepted by the Computational Neuroscience community.

​

Neymotin SA, Dura-Bernal S, Lakatos P, Sanger TD, Lytton WW. Multitarget multiscale simulation for pharmacological treatment of dystonia in motor cortex. Frontiers in Pharmacology 2016 7:157.

Example:

Phenomenological MSM: Virtual Brain (TVB) epileptor

1. context clear?

Epilepsy in a specific group of patients.

2. appropriate data?

Data from individual patient tractography and ECoG.

3. within a context?

Comparisons with clinical results -- seizures

4. limitations explicit?

Lack of explicit connections to cellular, metabolic or pharmacological

scales

5. version control

TVB is itself under git version control.

6. documented?

yes

7. disseminated?

papers, workshops, talks, posters

8. independent reviews?

paper reviewed

9. competing implementations?

unsure but the relative simplicity of model means could be coded in other ODE simulators

10. conformed to standards?

TVB itself provides a standard for simulation at this scale

​

Proix T, Bartolomei F, Guye M, Jirsa VK. (2017) Individual brain structure and modelling predict seizure propagation. Brain. 140:641-654.