SCENIC+ parameter modification for non-model organism

[nunezosc]

Hi,

Thank you for developing such an amazing tool @SeppeDeWinter @timoast . I am working in a nonstandard model organism and its genome size (< 0.5Gb) is much smaller than mice or humans, thus enhancers are not very common. Although I got the pipeline to work and generate some regulons (eRegulon_direct.tsv), I was wondering if the search space parameters can be modified to accommodate the smaller genome size. For instance, can the search space be 500-10000, or 100-5000? This will capture more promoter associated regulatory elements and reduce distal enhancers. However, will the use of a smaller search space have an impact on statistical validity since SCENIC+ was not originally created to infer regulons using promoter regions?

I also noticed that the Motif enrichment parameters (AUC, NES, and Rank threshold) may be too stringent for my analysis. Is there a recommended way to modify these thresholds to make them more lenient to obtain more regulons without compromising quality – e.g avoid false positives?

Additionally, the eRegulon_direct.tsv file has a column named “triplet_rank” which combines the scores of TF2Gene, TF2Region, and Region2Gene. How can these values be interpreted? – e.g are smaller triplet_rank values (closer to 0) more biologically meaningful than larger values?

Thank you!
Oscar