Pairwise Distances from Sequences

dist.hamming, dist.ml and dist.logDet compute pairwise distances for an object of class phyDat. dist.ml uses DNA / AA sequences to compute distances under different substitution models.

Usage

dist.hamming(x, ratio = TRUE, exclude = "none")

dist.ml(x, model = "JC69", exclude = "none", bf = NULL, Q = NULL,
  k = 1L, shape = 1, ...)

dist.logDet(x)

Arguments

x: An object of class phyDat
ratio: Compute uncorrected ('p') distance or character difference.
exclude: One of "none", "all", "pairwise" indicating whether to delete the sites with gaps, missing data (or ambiguous states). See details below.
model: One of "JC69", "F81" or one of 17 amino acid models see details.
bf: A vector of base frequencies.
Q: A vector containing the lower triangular part of the rate matrix.
k: Number of intervals of the discrete gamma distribution.
shape: Shape parameter of the gamma distribution.
...: Further arguments passed to or from other methods.

Value

an object of class dist

Details

So far 17 amino acid models are supported ("WAG", "JTT", "LG", "Dayhoff", "cpREV", "mtmam", "mtArt", "MtZoa", "mtREV24", "VT","RtREV", "HIVw", "HIVb", "FLU", "Blosum62", "Dayhoff_DCMut" and "JTT_DCMut") and additional rate matrices and frequencies can be supplied.

The "F81" model uses empirical base frequencies, the "JC69" equal base frequencies. This is even the case if the data are not nucleotides.

The argument exclude decides how gaps / ambiguous data / missing data are treated. Usually gaps are treated as ambiguous states, but you can give gaps its on state gap_as_state. exclude="none" keeps all ambiguous data. The behavior of dist.ml is in this case these same you would achieve using optim.pml to compute pairwise distances, it might be a bit odd. exclude="all" removes all sites with ambiguous states and all gaps if these are coded as ambiguous states. This can lead to the situation that there only few sites if any fo the alignment left. Safer is therefore to use exclude="pairwise" which only removes sites which are ambiguous for each pair of sequences.

References

Lockhart, P. J., Steel, M. A., Hendy, M. D. and Penny, D. (1994) Recovering evolutionary trees under a more realistic model of sequence evolution. Molecular Biology and Evolution, 11, 605–602.

Jukes TH and Cantor CR (1969). Evolution of Protein Molecules. New York: Academic Press. 21–132.

McGuire, G., Prentice, M. J. and Wright, F. (1999). Improved error bounds for genetic distances from DNA sequences. Biometrics, 55, 1064–1070.

Author

Klaus Schliep klaus.schliep@gmail.com

Examples


data(Laurasiatherian)
dm1 <- dist.hamming(Laurasiatherian)
tree1 <- NJ(dm1)
dm2 <- dist.logDet(Laurasiatherian)
tree2 <- NJ(dm2)
treedist(tree1,tree2)
#>      symmetric.difference   branch.score.difference           path.difference 
#>                4.00000000                0.05705091               30.95157508 
#> quadratic.path.difference 
#>                0.80097967 
# JC model
dm3 <- dist.ml(Laurasiatherian)
tree3 <- NJ(dm3)
treedist(tree1,tree3)
#>      symmetric.difference   branch.score.difference           path.difference 
#>                 6.0000000                 0.0412520                30.3644529 
#> quadratic.path.difference 
#>                 0.6106899 
# F81 + Gamma
dm4 <- dist.ml(Laurasiatherian, model="F81", k=4, shape=.4)
tree4 <- NJ(dm4)
treedist(tree1,tree4)
#>      symmetric.difference   branch.score.difference           path.difference 
#>                12.0000000                 0.1356107                40.7676342 
#> quadratic.path.difference 
#>                 2.0709714 
treedist(tree3,tree4)
#>      symmetric.difference   branch.score.difference           path.difference 
#>                8.00000000                0.09494752               39.52214569 
#> quadratic.path.difference 
#>                1.46345381