search

Teichlab / cellphonedb

MIT License
342 stars 105 forks source link

Prioritize results #135

Closed keenhl closed 4 years ago

keenhl commented 4 years ago

Hi all,

I ran the cellphoneDB analysis with the default parameters and there were lot of interactions with p-value equal to 0. My question is how do I further prioritize this list. Should I do more than 1000 permutations so that I can get p-values with greater precision or should I use the means values to sort the list of significant interactions.

Any thoughts.

Thank you very much.

mief commented 4 years ago

Hi, For prioritising interactions, you can use the significant_means table where the interactions are ranked based on the number of significant p-values. Another possibility is to be stricter with the expression level threshold and set it to 20% or higher.

Best, Mirjana