Talk:VERTEX

Comments by Thomas Kreuz

Overall, the article is fine and quite ready but I do have one major and a few minor points that should be addressed (all concerned with style and structure, for more detailed comments on the content please suggest an appropriate second reviewer):

General and only major issue:

If possible at the beginning please provide a bit more information about both the motivation behind VERTEX and the relevance of VERTEX for the potential user. I also would like to learn more about potential applications (this could also be a short section at the end). Together this would make the article read more like an introduction to a new software package (potentially motivating the reader to try it for him-/herself) and a bit less like a user’s manual. The current main section 2 (which does follow a very manual-like style) is fine as it is but adding some more general introductory and closing sections would decrease its relative weight in the article and that would be a good thing.

More minor issues:

The content of Fig.1A (basically the workflow from model to simulation to analysis plus some more details) does not really appear anyway early in the text. It can somehow be deduced from the overall structure of the article but it would be better if it were stated more explicitly right in the beginning, so before Section 2 which covers the contents of Fig. 1B. This could be some of the earlier general content mentioned above.

Then please also refer to the Figure 1 (A and B) in the main text.

Figure 3 is quite small and hard to read. In the zoom the subplot labels seem to be cut. Also please add a reference to that Figure in the text. References to Figure 2 and 4 are already there.

Please update citation 2 (in case the status has changed).

Reference 4 appears before reference 3 in the text.

Section 5 (5 Comparison with similar software tools) contains some slightly hidden criticisms of other software tools. Since this is supposed to be an Encyclopedia entry written from an impartial point of view please try to hide them a bit more (for example by stressing the advantages of VERTEX instead of the disadvantages of these tools).

For the sake of efficiency, all the (mostly minor) changes of the text I implemented immediately. Please use the history function to check these (and revert any that you do not agree with, in your reply then please add a small comment). Among others, for a more formal encyclopedia-like style I eliminated all ‘we’ (sometimes using ‘the user’ instead).

Reply to Thomas Kreuz

We have added a motivation section at the beginning of the article, which discusses in broad terms why we believe there is a need for a tool like VERTEX. We have also added a potential applications section at the end of the article which gives the reader an idea of how VERTEX may be used by discussing two examples of its use in simulating gamma oscillations and stimulus evoked activity.

We have discussed figure 1 (A and B) at the beginning of the section: What comprises a VERTEX simulation.

Figure 3 is now referenced in the text.

References 3 and 4 have been swapped around.

Section 5 has been updated to remove criticisms of other tools.

Comments by Hans Ekkehard Plesser

First of all, I wonder about the integration of VERTEX with the general simulation ecosystem. Your network construction example shows code explicitly setting compartment sizes and locations. Would't one usually want to load such data from tracing data provided in any of a number of standard formats? Does VERTEX support this? How about loading models described in standard languages such as NeuroML? The article should at least be clear about what is feasible and what not. The same holds for output data: is it only available as mat-files in a VERTEX internal format, or do you support export/conversion to common formats such as NEO?

I also find your section comparing VERTEX to other tools very short and just descriptive, without any information on the relative strengths of different tools. I noticed that Thomas Kreuz asked you to tone down criticism of other tools. I agree with his criticism, especially considering that you in the version of 7 May 2019 stated that Neuron "is designed for simulating single neurons or small networks in detail", which is rather off the mark given that Neuron is used today by, e.g., the Human and the Blue Brain Projects to simulate some of the largest and most complex cortical and non-cortical network models. But I think that you have gone too far in the opposite direction in your current version. A table comparing features might be useful, in the spirit of the comparisons provided in the Brette et al (2007) simulator review.

Finally, in Fig 4 you may also want to specify the number of compartments per model.

Reply to Hans Ekkehard Plesser

The Neuron Parameters section has been expanded to address your point about integration with the general simulation ecosystem. We mention that while VERTEX can load neuron geometries from NML files, it typically uses a caricature of the cell type with a limited number of compartments to reduce simulation size. In the Analysing simulation output section we now state that VERTEX saves its output as .mat file and that the user would be required to export data to their required format themselves. I think adding support for exporting to the neo format would be a great future project thought!

The comparison section has been updated to include a table and provide a bit more discussion.