Percent Identity Filter

Identify and filter subsets of sequences at a given sequence identity cutoff.

filter.identity(aln = NULL, ide = NULL, cutoff = 0.6, verbose = TRUE, ...)

Arguments

aln	sequence alignment list, obtained from seqaln or read.fasta, or an alignment character matrix. Not used if ‘ide’ is given.
ide	an optional identity matrix obtained from seqidentity.
cutoff	a numeric identity cutoff value ranging between 0 and 1.
verbose	logical, if TRUE print details of the clustering process.
...	additional arguments passed to and from functions.

Details

This function performs hierarchical cluster analysis of a given sequence identity matrix ‘ide’, or the identity matrix calculated from a given alignment ‘aln’, to identify sequences that fall below a given identity cutoff value ‘cutoff’.

Value

Returns a list object with components:

ind

indices of the sequences below the cutoff value.

tree

an object of class "hclust", which describes the tree produced by the clustering process.

ide

a numeric matrix with all pairwise identity values.

References

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

Author

Barry Grant

See also

read.fasta, seqaln, seqidentity, entropy, consensus

Examples

attach(kinesin)

ide.mat <- seqidentity(pdbs)

# Histogram of pairwise identity values
op <- par(no.readonly=TRUE)
par(mfrow=c(2,1))
hist(ide.mat[upper.tri(ide.mat)], breaks=30,xlim=c(0,1),
     main="Sequence Identity", xlab="Identity")

k <- filter.identity(ide=ide.mat, cutoff=0.6)
#> filter.identity(): N clusters @ cutoff =  10 
ide.cut <- seqidentity(pdbs$ali[k$ind,])
hist(ide.cut[upper.tri(ide.cut)], breaks=10, xlim=c(0,1),
     main="Sequence Identity", xlab="Identity")

#plot(k$tree, axes = FALSE, ylab="Sequence Identity")
#print(k$ind) # selected
par(op)
detach(kinesin)