Kamada Kawai vs Fruchterman Reingold I am trying to visualize microbial correlation networks. I am using igraph in R. And my goal is to have weighed edges based on the correlation coefficient between the nodes.
I don't understand what the difference is between the Kamada Kawai and the Fruchterman Reignold algorithms. I can have weighed edges using either of them. Also you might have some advice on which one to use in my case. 
 A: Both of these layout algorithms belong to the family of force-directed layout methods. You can think of these as applying a force to each vertex depending on their connections, moving the vertices based on these forces (or rather: minimizing the energy of the system), though I have to say that the choice of forces is often thoroughly non-physical.
igraph's implementation of these two methods follows the original papers closely for the case of connected graphs. You can read more about the details here:

*

*https://igraph.org/c/html/latest/igraph-Layout.html#igraph_layout_fruchterman_reingold

*https://igraph.org/c/html/latest/igraph-Layout.html#igraph_layout_kamada_kawai
The Kamada-Kawai method inserts a spring between all pairs of vertices. The length of the spring is set to be the same as the graph distance between vertices. This means that edges with a large weight will be longer. The KK layout does particularly well will lattice-like networks.
The Fruchterman-Reingold method uses an attractive force between vertices that are directly connected with an edge, and a repulsive force between all vertex pairs, whether connected or not. The attractive force is proportional to the edge's weight, thus edges with a large weight will be shorter. The FR layout is a good baseline choice for most types of networks.
You can try both, and choose the visualization which seems clearer. Remember that you should not read too much into a graph visualization.  Generally, it is not possible to lay out any network so that connected vertices are near each other and non-connected ones are further apart. Visualizations are useful for a first look, but they are not suitable for drawing hard conclusions about the structure of your network.
