soma positions are available in original dataset

[jefferis]

I just picked up your eLife paper with interest. One point though. I was surprised to note in the methods that the mouse retina soma positions were manually extracted. When I processed these data

https://github.com/jefferis/nat.examples/tree/master/03-helmstaedter2013

there was a structure that contained the soma coordinates within this matlab file:

http://neuro.rzg.mpg.de/download/Helmstaedter_et_al_Nature_2013_skeletons_contacts_matrices.mat

linked from their accompanying website http://www.neuro.mpg.de/connectomics. Did this not do the trick for you?

Best wishes,

Greg.