Data-driven axonal guidance

[mindbound]

Would it be possible to implement white matter/axonal guidance using e.g. DSI Studio, in order to model realistic connectivity between distant brain regions? Given that VERTEX already supports compartmental models, I think this should be doable but there may be difficulties I'm not aware of. TIA.

[richardtomsett]

Apologies for the delay, github didn't notify me about your message! This would be possible, but require some fairly detailed modifications. The current implementation was made with modelling local networks, with in vitro slices particularly in mind. You would need to add functions to implement rules that convert large-scale gross white matter anatomy to the individual neuron level, and also functionality to deal with multiple axonal transmission speeds for local or distant neurons (unmyelinated/myelinated). The connectivity data structure is just a big sparse matrix so this isn't a barrier, but you're right there could be quite a few implementation details that would make this a non-trivial project.

I'm aiming to write some better documentation for VERTEX's internals over the next few months so that others will hopefully find it easier to make extensions, but I don't think VERTEX is best suited for the kind of modifications you're suggesting at the moment, given its local-network focus.

[mindbound]

Thank you, that's more or less how I imagined the situation. In any case, having documentation w.r.t. extending the code will be great.