Align communities from two or more networks

Find equivalent communities from two or more networks and re-assign colors to them in a consistent way across networks. A ‘new.membership’ vector is also generated for each network, which maps nodes to community IDs that are renumbered according to the community equivalency.

community.aln(x, ..., aln = NULL)

Arguments

x, ...	two or more objects of class cna (if the numbers of nodes are different, an alignment ‘fasta’ object is required for the aln argument; See below) as obtained from function cna. Alternatively, a list of cna objects can be given to x.
aln	alignment for comparing networks with different numbers of nodes.

Value

Returns a list of updated cna objects.

Details

This function facilitates the inspection on the variance of the community partition in a group of similar networks. The original community numbering (and so the colors of communities in the output of plot.cna and vmd.cna) can be inconsistent across networks, i.e. equivalent communities may display different colors, impeding network comparison. The function calculates the dissimilarity between all communities and clusters communities with ‘hclust’ funciton. In each cluster, 0 or 1 community per network is included. The color attribute of communities is then re-assigned according to the clusters through all networks. In addition, a ‘new.membership’ vector is generated for each network, which mapps nodes to new community IDs that are numbered consistently across networks.

See also

cna, plot.cna, vmd.cna

Examples

# \donttest{
  # Needs MUSCLE installed - testing excluded
  if(check.utility("muscle")) {

    if (!requireNamespace("igraph", quietly = TRUE)) {
      message('Need igraph installed to run this example')
    } else {

    ## Fetch PDB files and split to chain A only PDB files
    ids <- c("1tnd_A", "1tag_A")
    files <- get.pdb(ids, split = TRUE, path = tempdir())
    
    ## Sequence Alignement
    pdbs <- pdbaln(files, outfile = tempfile())
    
    ## Normal mode analysis on aligned data
    modes <- nma(pdbs, rm.gaps=TRUE)
    
    ## Dynamic Cross Correlation Matrix
    cijs <- dccm(modes)$all.dccm
 
    ## Correlation Network
    nets <- cna(cijs, cutoff.cij=0.3)

    ## Align network communities
    nets.aln <- community.aln(nets)

    ## plot all-residue and coarse-grained (community) networks
    pdb <- pdbs2pdb(pdbs, inds=1, rm.gaps=TRUE)[[1]]
    op <- par(no.readonly=TRUE)

    # before alignment
    par(mar=c(0.1, 0.1, 0.1, 0.1), mfrow=c(2,2))
    invisible( lapply(nets, function(x) 
       plot(x, layout=layout.cna(x, pdb=pdb, k=3, full=TRUE)[, 1:2], 
               full=TRUE)) )
    invisible( lapply(nets, function(x) 
       plot(x, layout=layout.cna(x, pdb=pdb, k=3)[, 1:2])) )

    # after alignment
    par(mar=c(0.1, 0.1, 0.1, 0.1), mfrow=c(2,2))
    invisible( lapply(nets.aln, function(x) 
       plot(x, layout=layout.cna(x, pdb=pdb, k=3, full=TRUE)[, 1:2], 
               full=TRUE)) )
    invisible( lapply(nets.aln, function(x) 
       plot(x, layout=layout.cna(x, pdb=pdb, k=3)[, 1:2])) )

    par(op)     

    }
  }
#> Warning: /var/folders/xf/qznxnpf91vb1wm4xwgnbt0xr0000gn/T//Rtmp4WslmZ/1tnd.pdb  exists. Skipping download
#> 
  |                                                                            
  |                                                                      |   0%
  |                                                                            
  |===================================                                   |  50%
  |                                                                            
  |======================================================================| 100%
#> Reading PDB files:
#> /var/folders/xf/qznxnpf91vb1wm4xwgnbt0xr0000gn/T//Rtmp4WslmZ/split_chain/1tnd_A.pdb
#> /var/folders/xf/qznxnpf91vb1wm4xwgnbt0xr0000gn/T//Rtmp4WslmZ/split_chain/1tag_A.pdb
#> ..
#> 
#> Extracting sequences
#> 
#> pdb/seq: 1   name: /var/folders/xf/qznxnpf91vb1wm4xwgnbt0xr0000gn/T//Rtmp4WslmZ/split_chain/1tnd_A.pdb 
#> pdb/seq: 2   name: /var/folders/xf/qznxnpf91vb1wm4xwgnbt0xr0000gn/T//Rtmp4WslmZ/split_chain/1tag_A.pdb 
#> 
#> Details of Scheduled Calculation:
#>   ... 2 input structures 
#>   ... storing 936 eigenvectors for each structure 
#>   ... dimension of x$U.subspace: ( 942x936x2 )
#>   ... coordinate superposition prior to NM calculation 
#>   ... aligned eigenvectors (gap containing positions removed)  
#>   ... estimated memory usage of final 'eNMA' object: 13.5 Mb 
#> 
#> 
  |                                                                            
  |                                                                      |   0%
  |                                                                            
  |===================================                                   |  50%
  |                                                                            
  |======================================================================| 100%
# }