search

stefpeschel / NetCoMi

Network construction, analysis, and comparison for microbial compositional data
https://netcomi.de
GNU General Public License v3.0
154 stars 28 forks source link

input for net construct() #132

Open nandhu1705 opened 16 hours ago

nandhu1705 commented 16 hours ago

Hi, Just a basic question before I start the analysis. netConstruct would need the raw counts for analysis right?. Then it apply CLR normalization and it would take the normalized counts for the network construction right?

after CLR normalization, will netConstrust convert it to relative abudance or will it keep the CLR normalized counts to build the network analysis. ? net_pears <- netConstruct(amgut2.filt.phy,
measure = "pearson", normMethod = "clr", zeroMethod = "multRepl", sparsMethod = "threshold", thresh = 0.3, verbose = 3)

stefpeschel commented 11 hours ago

Yes, exactly. Here is an illustration of the workflow performed inside netConstruct(), only that you would use multiplicative replacement instead of pseudo counts.

Neither relative abundances nor clr transformed counts are used for network analysis, because the clr transformed counts are used to estimate correlations. These correlations are then transformed into similarity values / edge weights on which the network analysis and plot are based.