search

Teichlab / cellphonedb

MIT License
340 stars 105 forks source link

Number of interactions in count_network.tsv #187

Closed TobiTekath closed 3 years ago

TobiTekath commented 4 years ago

Hi,

I just noticed that when plotting the heatmap the number of interactions in the "count_network.tsv" ("--count-network-name") are not correct, as the (significant) interactions of A -> B and B -> A are summed together beforehand (if A and B are different cell types).

Or is there a specific reasoning for this behaviour?

The same goes for the actual plot, which explains the perfect symmetry of the plot.

panxiaoguang commented 4 years ago

you can see here https://github.com/Teichlab/cellphonedb/blob/master/Docs/RESULTS-DOCUMENTATION.md

TobiTekath commented 4 years ago

Thanks @panxiaoguang for the link, but what I meant to point out is that it is counter intuitive to write out a table of the number of interactions which separate entries for A -> B and B -> A, but both having the same number because they both represent A <-> B.

I hope that made my point more clear. Thanks.