Figure 4: SCNIC Feature Reduction and Visualization of SCNIC
Networks
A) We used SCNIC to select modules using the OTU table from the HIV
dataset, the SMD module selection algorithm, and SparCC R-values ranging
from 0.2 to 1.0, in increments of 0.05. The R-value is plotted against
the number of features in the resulting BIOM table produced by SCNIC. As
the R-value increases the number of modules decreases and the number of
single features (modules + OTUs not included in modules) increases.
After the R-value of 0.65, the number of features in the resulting file
remained the same at 4351 features, because there were no modules that
were created past a SparCC R of 0.65.The Cytoscape output allows for
easy exploration and visualization of relationships between different
OTUS/taxa in an interactive interface. B) R = 0.65 C) R = 0.4 D) R =
0.2. As the R-value increases, the size of the network decreases as
SCNIC does not include singletons (features with no significant positive
correlations) in the resulting network file. Correlation network where
edges are correlations with a R-value greater than the threshold set.
Nodes are OTUs and node color represents module membership (i.e.
module-0 is pink in Panel B).