grownet · network viewer

Read a grownet interaction network: focus by species, inspect the evidence behind each edge, download for Cytoscape or the pipeline.

How a network is derived

This viewer reads a network that is already derived; it does not derive. The settings live on the local page (grownet gui, Advanced settings) and on the command line (grownet derive --help), with the same defaults, and every network records the ones it was made with in meta.settings. By default: the area under the curve, compared as the mean log2 over replicate sets, after a no-growth check; strains keyed by NCBI taxon id; two-member co-cultures and drop-out communities; the same species-identifying technique on both sides; batch cultures only; an interaction absent when |log2 mean| is under one sd; the test the derivation ran, named in the network's meta.statistics, with Benjamini-Hochberg, reported and never deciding. The reasons are in docs/METHOD_NOTES.md.

Filter this view (live)

Minimum |interaction strength|: 0.0

Show effects

Absence threshold k: 1.0

A comparison reads absent when |log2 mean| is under k standard deviations of its own spread. Moving k retunes the view; it does not rederive.

Show absent comparisons (tested, no interaction)

Off by default, as the pipeline and the Cytoscape style do. An absence is a measured result, not missing data.

Show compromised edges (pooled strains, detected removed member, non-batch)

Off by default, as the pipeline exports them. A single-replicate edge is not in this group: it is shown and marked, because the comparison is sound and only its spread is unknown.

Minimum supporting studies

Counts the studies each arc cites. Above 1 it needs a network derived with merged arcs (Merge parallel arcs, or --merge-arcs), since an arc as derived rests on one study.

Defaults follow docs/METHOD_NOTES.md. The derivation choices are a scientific decision made by the collaboration; only the live filters on the right change the network already loaded.