Scoring matrices quantify the log2 odds ratio that a segment pair with a given distance and absolute dot product come from a pair of neurons of the same type, rather than unrelated neurons.
These scoring matrices were generated using all by all pairs from 150 DL2 antennal lobe projection neurons from the http://flycircuit.tw dataset and 5000 random pairs from the same dataset.
smat.fcwb
was trained using nearest-neighbour distance and the
tangent vector defined by the first eigen vector of the k=5 nearest
neighbours.
smat_alpha.fcwb
was defined as for smat.fcwb
but weighted
by the factor alpha
defined as (l1-l2)/(l1+l2+l3) where l1,l2,l3 are
the three eigen values.
Most work on the flyircuit dataset has been carried out using the
smat.fcwb
scoring matrix although the smat_alpha.fcwb
matrix
which emphasises the significance of matches between linear regions of the
neuron (such as axons) may have some advantages.
Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.
Please suggest features or report bugs with the GitHub issue tracker.
All documentation is copyright its authors; we didn't write any of that.