Persistence diagram failed/empty list for some neurons

[bluebrain]

Persistence diagram failed/empty list for some neurons

[eleftherioszisis]

Hi and thanks for opening an issue. Could you please provide an SWC file example to reproduce the error?

[yanyanming77]

Hi:

Thanks for the response!

I've figured out the 0-length issue (some neurites are stored as type=0 in the SWCs so they are not read correctly when I specify dendrite_type = 'basal_dendrite'), so we're good with that. But unfortunately there are still some files that have "key error" issue.

Please see the attached example. I converted it to .txt format since .swc is not supported by Github, you may need to change it back.

By looking into this example, the get_ph_neuron() method first get a dictionary called ‘parents’, which is a dictionary of {children:parent} pairs, and use birth and death time to get the p-diagram, after the first round of iteration over the active nodes, a parent node is added to the active list, which is not a key in the ‘parents’ dictionary, thus throw the key error.

For example, for the 5th dendrite:

1. The parent dictionary of the 5th dendrite extracted is: {1: 0, 4: 1, 10: 4, 33: 10, 92: 10, 131: 4, 139: 1, 195: 139, 202: 140}
2. The active nodes in the first iteration are: [33, 92, 131, 195, 202], they all have keys in the parent dictionary
3. The active nodes in the second iteration are: [4, 139, 140], while 140 is not a key in the parent dictionary, which creates the error

Please let me know if you have any thoughts on this. Thank you very much!

Best,
Ming
example_swc_failed_getPH_neuron.txt

[lidakanari]

Hi Ming,

Thanks for reporting this issue.

I checked your data and I see where it breaks. It looks like one of the sections connects to the wrong place.

I suggest that you use the following way to load your files and I will check again later what goes wrong in the main loader.

neuron = tmd.io.load_neuron_from_morphio('example_swc_failed_getPH_neuron.swc')

This extracts the persistence diagrams correctly