Help understanding causal network output

[ebasto]

Hello,

I wanted to ask for clarification regarding the output of the causal network vignette "Hypothesis-testing for CCC & Downstream Signalling Networks".

Here is my current understanding:
column name (observed values when only a few) - description
source - source node in the PPI, there are some values here that I don't understand ("_s","_pert_c0","_meas_c0","")
source_type (unmeasured,input) - are the input nodes the receptors that were identified in the previous step? are unmeasured genes that have no expression? or genes that are not part of that receptor group?
source_weight - not sure what this is, the example only has two values (0 and -4.66)
source_pred_val (1,0,-1) - node is upregulated, downregulated, is zero no differential expression?
target - target node in the PPI, there are some values here that I don't understand ("_s","_t","_meas_c0","")
target_type (unmeasured, output) - similar to "source_type" except "output" would be TFs?
target_weight - not sure what this is? the example has a few more values than "source_weight", they are all 0 or positive.
target_pred_val (1,-1) - node is upregulated, downregulated? no zeros in this one
edge_type (1,-1,0) - unsure what this is
edge_pred_val(1,-1) - whether the edge upregulates or downregulates the target node.

Thank you in advance!

Edgar

[dbdimitrov]

Hi @ebasto,

These ("_s","_t","_meas_c0","") are just dummy variables used when setting the ILP problem to be solved.

source_weight and target_weight are the inputs provided to the causal net method (the "measured" nodes).

edge type is the type (sign) of interaction in the prior knowledge network, i.e. it is not predicted by the method.
e.g, A -> B, the edge_type is 1. If A=-1, source_pred_val=-1, edge_pred_val=-1, and target_pred_val=-1;

Unmeasured nodes would be those that are not input or output - I.e. all other nodes (except the provided receptors or TFs in the tutorial).

All the rest you got correct :)

Thanks for pointing this out also. I realize that maybe it's a bit of a black box how it works.

I will keep this issue open for now until I provide a bit more details in the tutorial.