Principal Component Analysis

Performs principal components analysis (PCA) on torsion angle data.

# S3 method for tor
pca(data, ...)

Arguments

data

numeric matrix of torsion angles with a row per structure.
...

additional arguments passed to the method pca.xyz.

Value

Returns a list with the following components:

L

eigenvalues.

U

eigenvectors (i.e. the variable loadings).

z.u

scores of the supplied data on the pcs.

sdev

the standard deviations of the pcs.

mean

the means that were subtracted.

References

Grant, B.J. et al. (2006) Bioinformatics 22, 2695--2696.

Author

Barry Grant and Karim ElSawy

See also

torsion.xyz, plot.pca, plot.pca.loadings, pca.xyz

Examples

##-- PCA on torsion data for multiple PDBs 
attach(kinesin)

gaps.pos <- gap.inspect(pdbs$xyz)
tor <- t(apply( pdbs$xyz[, gaps.pos$f.inds], 1, torsion.xyz, atm.inc=1))
pc.tor <- pca.tor(tor[,-c(1,233,234,235)])
#plot(pc.tor)
plot.pca.loadings(pc.tor)


detach(kinesin)

if (FALSE) {
##-- PCA on torsion data from an MD trajectory
trj <- read.dcd( system.file("examples/hivp.dcd", package="bio3d") )
tor <- t(apply(trj, 1, torsion.xyz, atm.inc=1))
gaps <- gap.inspect(tor)
pc.tor <- pca.tor(tor[,gaps$f.inds])
plot.pca.loadings(pc.tor)
}