why the n_conns in LabelConnectivity becomes 0 while in VertexConnectivity is not?

[Zhouyu0206]

Hi,
Thanks a lot for creating such a handy package!
I got a VertexConnectivity object 'con_clust' which shows that '<VertexConnectivity | n_sources=3331, n_conns=329, subject=fsaverage>', when I tried to parcel it using 'con_parc = con_clust.parcellate(labels, summary='degree', weight_by_degree=False)', the info of con_parc shows that: <LabelConnectivity | n_sources=68, n_conns=0, subject=fsaverage>.
I'm wondering why the n_conns in LabelConnectivity becomes 0 while in VertexConnectivity is not?

[wmvanvliet]

Good question... could you share your VertexConnectivity file somehow? (Feel free to send me an email: marijn.vanvliet@aalto.fi). This sounds like something that should never happen.

[Zhouyu0206]

Hi! Marijn,
I've already sent it to your email. Thank you so much!

[wmvanvliet]

Ah... it's because the connection strength for every connection in your VertexConnectivity object is 0. Double-check your connectivity metric. If you don't care about connection strength, try setting it to all ones:

con.data = np.ones_likes(con.data)

[wmvanvliet]

A connection with a strength of 0 is actually treated by conpy as no connection at all

[Zhouyu0206]

I see, thank you very much! But if the strength of every connection in my VertexConnectivity object is 0, why they existed in the first place? I mean if the strength of a connection is 0, why this connection would exist in the VertexConnectivity object?