bio3d (version 2.3-4)

plot.matrix.loadings: Plot Residue-Residue Matrix Loadings

Description

Plot residue-residue matrix loadings of a particular PC that is obtained from a principal component analysis (PCA) of cross-correlation or distance matrices.

Usage

# S3 method for matrix.loadings
plot(x, pc = 1, resno = NULL, sse = NULL,
  mask.n = 0, plot = TRUE, ...)

Arguments

x

the results of PCA as obtained from pca.array.

pc

the principal component along which the loadings will be shown.

resno

numerical vector or ‘pdb’ object as obtained from read.pdb to show residue number on the x- and y-axis.

sse

a ‘sse’ object as obtained from dssp or stride, or a ‘pdb’ object as obtained from read.pdb to show secondary structural elements along x- and y-axis.

mask.n

the number of elements from the diagonal to be masked from output.

plot

logical, if FALSE no plot will be shown.

...

additional arguments passed to plot.dccm.

Value

Plot and also returns a numeric matrix containing the loadings.

Details

The function plots loadings (the eigenvectors) of PCA performed on a set of matrices such as distance matrices from an ensemble of crystallographic structures and residue-residue cross-correlations or covariance matrices derived from ensemble NMA or MD simulation replicates (See pca.array for detail). Loadings are displayed as a matrix with dimension the same as the input matrices of the PCA. Each element of loadings represents the proportion that the corresponding residue pair contributes to the variance in a particular PC. The plot can be used to identify key regions that best explain the variance of underlying matrices.

References

Skjaerven, L. et al. (2014) BMC Bioinformatics 15, 399. Grant, B.J. et al. (2006) Bioinformatics 22, 2695--2696.

See Also

plot.dccm, pca.array

Examples

# NOT RUN {
   attach(transducin)
   gaps.res <- gap.inspect(pdbs$ali)
   sse <- pdbs$sse[1, gaps.res$f.inds]

   # calculate modes
   modes <- nma(pdbs, ncore=NULL)

   # calculate cross-correlation matrices from the modes
   cijs <- dccm(modes, ncore=NULL)$all.dccm

   # do PCA on cross-correlation matrices
   pc <- pca.array(cijs)

   # plot loadings
   l <- plot.matrix.loadings(pc, sse=sse)
   l[1:10, 1:10]

   # plot loadings with elements 10-residue separated from diagonal masked
   plot.matrix.loadings(pc, sse=sse, mask.n=10)

# }