Figure 1: SCNIC Schematic and Data Flow
A. The basic process of SCNIC involves first identifying pairwise
correlations between species and using them to build a correlation
network. Modules of correlated features are identified and then
summarized for downstream statistical analysis, or multi-omic analysis
between modules of microbes and other feature types. B. The input to
SCNIC comes in the form of a count table in BIOM format. The first step
takes the table and generates a correlation table and network. The table
is in a tab delimited format and the network is in GML format and can be
used to visualize the network in Cytoscape. Modules are detected and
summarized in the final step which generates a module membership file
indicating which features are in each module. The collapsed BIOM table
contains the same total counts per sample as the original table, but
with less features. All features not included in modules are retained
with their original counts and all modules have a total count per sample
of the sum of the counts of all features in that module.