NEURON vs GENESIS - Opinions, please?

Matt Jones jonesmat at
Mon Jun 24 11:21:50 EST 2002

mats_trash at (mat) wrote in message news:<43525ce3.0206240206.32438f65 at>...
> > Not that anyone cares, but I've decided on GENESIS.
> I was also wondering whether to opt for genesis or neuron.  Certainly
> Neuron seems to have been used in more publications, but almost all
> seem to be studies of single neurones rather than networks which I
> would be more interested in.  As a side-project I'm currently learning
> Fortran and (attempting) to program my own simulator, which hopefully
> should give me much greater control and also slim down the software so
> that it runs faster.
> Why did you decide on Genesis?
> How have your initial 'experiments' gone?
> Cheers

Wow, someone did actually care....

Here's an interesting piece of trivia: 
Question: What is the most popular compartmental modeling software in
the literature?
Answer: "Other" - (i.e., neither Neuron nor Genesis, but usually a
custom-written program for the purpose of running just that one

I decided on Genesis for 2 reasons:

1) There is actually documentation available for Genesis, and a fairly
active user maillist, an actual textbook that guides you through its
use, etc. These exist for Neuron too. The user maillist is probably
even more active. But the "documentation" available for Neuron is
pretty disorganized and sparse. Also, my experience with Neuron is
that the user interface is pretty clunky, and in a previous
incarnation of Neuron (for Mac) I found it impossible to properly save
a simulation and re-run it later with the same settings. I'm sure
that's been fixed by now, but it was enough to put me off. Don't know
about Genesis' interface yet.

2) Genesis is built to work like unix, which almost certainly means it
is extremely flexible and expandable (and indeed, the designers had
large scale network simulations in mind). I've been using other large
scale unix simulation software recently, and have found it incredibly
useful to be able to write a script (in Matlab) that writes a script
(in unix) that runs hundreds of separate simulations and organizes the
output, and then pipes it into an analysis script, etc. I literally
spend the afternoon setting up the scripts, then run them just before
I go home, and when I come in the next day the simulations have been
run, and the data have been analyzed according to my specifications
and the figures have been made (!). Well, ok, it's not quite as simple
as that. But working in unix has a lot of advantages for automating
various things.

Haven't actually run any Genesis simulations yet, but have ordered the
book and downloaded the binaries (precompiled for Mac OS X no less -
how perfect is that?) and demos.

Before you launch into coding your own simulator, I would recommend
having a look at the Genesis or Neuron code itself, because there are
apparently a number of useful tricks for making solving all those
differential equations run much more efficiently. At least this is
what I've been told.



More information about the Neur-sci mailing list