Hi, I am trying to run a network analysis for the first time on experimental data and have run into a few questions. My data is examining how the microbiome of an algal species is altered by environmental factors (nutrients and temperature). I have run six experiments with five treatments with water from two different lakes. I have run other analyses on this data such as PCoA and know there is ASV overlap between my experiments but the community structure differs, especially between lakes. There is also some partitioning between treatments within experiments but thus partitioning is not significant. My ultimate goal is to identify nodes that are hubs that remain important across treatments or are highly correlated to my algal species abundance, and to identify taxa in modules that differ between treatments. I am trying to use the SINC sparcc plugin for this analysis but am not sure how to partition the data. Specifically
I have seen network analysis used on time series data but have not found any examples for experimental data. Can you run network analysis on experimental data and if so do you have to create a separate network for each experiment, treatment, ect?
I originally tried running the sparcc correlations per experiment but kept getting an error that from other posts I determined was due to their not being enough samples/features for comparisons. each experiment has 12 or 15 samples and there are around 160 features (these have been collapsed at the family level). Since this does not work on a per experiment basis due to a too small sample size is there a way to run the correlations on the full data set and then filter by experiment/treatment after, prior to building the network?