I’m getting weird behavior in SCNIC for module analysis and I’m hoping you can help me. I’m generating 3 networks, one with all my samples and one stratified by a metadata category. I have a bad habit of pounding through an analysis in Jupyter and essentially re-running everything. My problem is that I’m running into unstable behavior. Nothing major in my environment has changed between runs: my tables are the same, my computer is the same, my conda installation has not changed. Yet, I’m getting instability in the network and modules.
I’m finding the that the names aren’t consistent, which is a pain in the butt as I’m trying trying to check for redundancy in the modules across the three networks. Is there some state variable I’m missing which would let me set the numbering etc. consistently? Has anyone else observed instability?