"NeuroPath2Path: Classification and elastic morphing between neuronal arbors using path-wise similarity" - Tamal Batabyal, Barry Condron, Scott T. Acton.

This tool computes morphological similarity between a pair of neurons, provides a visualization framework of contnuous morphing between neurons.

_________________________ NeuroPath2Path ________________________

1. Set the Directory of data.

Open neuroClassifyInterclass.m (similarly neuroClassifyIntraclass.m for intraclass classification). Set the newpath as the path where the directory containing the dataset is present.

2. Run jobClassifyInter.slurm (similarly jobClassifyIntra.slurm). The default parameter values are given. User can choose different parameters by opening the .slurm file and changing the arguments of neuroClassifyInterclass().

Ex: neuroClassifyInterclass(0.2,15,5,3,9) indicates that this instance of experiment tests on the 20% perclass-randomly sampled data, while rest of the data are used to represent all classes. As mentioned in the paper, we use the number of paths, which is a global feature, of a test neuron (from 20% data) to find the candidates of neurons (from 80% data) to perform majority voting. The binwidth of perclass-path distribution is 15. For a test neuron, the closest candidates are searched through step length 5 (search space reduction) and continue till +/- (95) paths from the number of paths of the test neuron. For example, if a test neuron has 60 rooted paths, we searched the possible candidates from 80% of data, which have numbers of paths ranging from 60-(95)=15 to 60+(9*5)=105. Majority vote is performed on 3 nearest neighbors.

For details, please follow the paper.

3. Once the job is completed, open OUTPUT.log

4. The python version is under preparation

% __________File descriptions

% classification readinput : Read swc files from directory computenP2P : Compute path features (User can add more features)

computeDSTInter & Intra : Path alignment + distance between paths

munkres + distanceNeuronNew : Distance between a pair of neurons and path correspondence

jobClassifyInter and jobClassifyIntra : job script files to run in cluster

neuroClassifyInterclass and neuroClassifyIntraclass: main files

% importance weights jobIW : jobscript to get the importance weights

getFeatureWeights : Obtain the importance weights