I have a data.frame with the correlations between OTUs and genes. I want to make a network in order to know what genes are related to what OTUs.

This is the head of my dataframe :

1 OTU0016 OTU0017 0.904
2 OTU0016 OTU0041 0.917
3 OTU0017 OTU0041 0.906
4 OTU0014 OTU0046 0.908
5 OTU0037 OTU0046 0.908
6 OTU0017 OTU0062 0.909

Should I only keep the correlations between OTU/gene , or do I need to stay with OTU/OTU + gene/gene + OTU/gene correlations? (Because it is a huge data.frame and if I only would keep the OTU/gene correlations, it will reduce the size)

Then, I construct the network with igraph. I extract the different components with split(names(V(my_graph)), components(my_graph)$membership) . That extracts correctly the components, but, the first one is compound of 361805 genes and OTUs, whereas the others have "correct" numbers of OTUs and gene (generally one OTU for 8 genes) . Is there a solution to reduce the size of the first components or does it look "normal" ?

Bests