SCENIC output

[saeedfc]

Hi @cflerin ,

There are two outputs in the standard pipeline.

1. Adjacencies based on GRN. Here we see an 'importance' score for each target gene and Transcription factor relationship.
2. Later after pruning, we get regulons. In this output we have weights for each gene for each MOTIF of each Transcription factor.

In several issues that has been discussing scenic outputs here in the 'issues' page (Issues 157 , 190 , 94 ) , I may have confused these two outputs. The 'importance' based on Adjacencies does not seem to be the same as the 'weight' in the regulons file. The values seem different. (Also for each motif for the same TF, you might have a single target gene repeating).

To make the best out of SCENIC output, I might want to see the set of targets of a transcription factor. Or query a set of genes to see whether they have a particular transcription factor associated with them. For some of these purposes, I would like to extract weights attached to a TF-Target relationship.

1. From your comments under issue 157 I assume you would suggest to pick the weights in the regulons file ( I still don't know where this value comes from) to evaluate the relationship of a Transcription Factor to its target gene. Is that correct? Is there a ready made function to extract a regulon (a transcription facto, its targets and weight) included in the pyscenic package or already available elsewhere as an R function that I did not find yet in the repository?

2. Also, consider a target gene X under Transcription Factor PPARG. For different motifs of PPARG, X has different weights. What would you suggest to consider to evaluate the relationship between PPARG and X (an average of weights or the highest one?)

3. When AUC matrix is created, how is the information from each motif (that belongs to same transcription factor) compressed to one value for a single Transcription factor (Regulon)? From the output it seams, each motif is considered as one gene set.

Thanks a lot for your time, Kind regards, Saeed

[withermatt]

Hi @saeedfc,

Have you come to any conclusions regarding your questions? I am wondering these myself.