### Modelling network without sufficient data

Posted:

**Fri Jul 24, 2015 8:29 am**Hi everyone

First off, I want to say that I have little to none programming knowledge and this is my first computational project. Therefore, I'd like to apologise if the answer I seek is common-sensical or can be easily found with some reading.

1) This has nothing to do with NEURON or the hoc code yet. I have attempted to construct a hypothetical network model from very little anatomical and functional data, and I am just wondering if someone could comment on whether what I've constructed makes sense. Say, we know there are both excitatory and inhibitory neurons in a certain brain area and we know that there are more than one of each kind. For the sake of discussion, I'll name them E1, E2, E3 (excitatory neuron types), I1 and I2 (inhibitory neuron types). I don't really know specifically how each of these neurons are connected to each other but I have some clues. For example, I know E1 receives both excitatory and inhibitory input while E2 only recieves excitatory input. Would it make sense then that I connect E2, E3, I1 and I2 as E1's presynaptic neurons, and E1 and E3 as E2's presynaptic neurons (because these are all possible connections)? Say I also know the types of input E3, I1 and I2 receive (inhibtory/excitatory). Would it make sense for me to do what I did with E1 and E2 to the rest of the neurons and make a model out of it? Apart from the opinion that it makes a good starting point (I would be able to remove non-existing connections later anyway), what else can I experiment with using such a model?

2) I've read some papers whereby a synaptic weight and divergence value have been given to a particular pair of neurons, and based on these connectivity parameters, neurons were "randomly" (I quote the paper) wired together. I want to implement this as well. It makes sense that while we know that A connects to B and B connects to C, not all A is connected to C via a disyanptic connection. A-B and B-C connections can exist separately from each other. How can this randomisation be efficiently implemented between simulation "trials"?

I'm not sure if I explained both of my problems properly. If you have any questions or comments please let me know. I'd like to hear from you! Thank you for your kind attention!

First off, I want to say that I have little to none programming knowledge and this is my first computational project. Therefore, I'd like to apologise if the answer I seek is common-sensical or can be easily found with some reading.

1) This has nothing to do with NEURON or the hoc code yet. I have attempted to construct a hypothetical network model from very little anatomical and functional data, and I am just wondering if someone could comment on whether what I've constructed makes sense. Say, we know there are both excitatory and inhibitory neurons in a certain brain area and we know that there are more than one of each kind. For the sake of discussion, I'll name them E1, E2, E3 (excitatory neuron types), I1 and I2 (inhibitory neuron types). I don't really know specifically how each of these neurons are connected to each other but I have some clues. For example, I know E1 receives both excitatory and inhibitory input while E2 only recieves excitatory input. Would it make sense then that I connect E2, E3, I1 and I2 as E1's presynaptic neurons, and E1 and E3 as E2's presynaptic neurons (because these are all possible connections)? Say I also know the types of input E3, I1 and I2 receive (inhibtory/excitatory). Would it make sense for me to do what I did with E1 and E2 to the rest of the neurons and make a model out of it? Apart from the opinion that it makes a good starting point (I would be able to remove non-existing connections later anyway), what else can I experiment with using such a model?

2) I've read some papers whereby a synaptic weight and divergence value have been given to a particular pair of neurons, and based on these connectivity parameters, neurons were "randomly" (I quote the paper) wired together. I want to implement this as well. It makes sense that while we know that A connects to B and B connects to C, not all A is connected to C via a disyanptic connection. A-B and B-C connections can exist separately from each other. How can this randomisation be efficiently implemented between simulation "trials"?

I'm not sure if I explained both of my problems properly. If you have any questions or comments please let me know. I'd like to hear from you! Thank you for your kind attention!