TKF (version 0.0.8)

TKF92: Evolutionary distance estimation with TKF92 model

Description

This function implements the TKF92 model to estimate the pairwise distance from protein sequences.

Usage

TKF92(fasta, mu=NULL, r=NULL, expectedLength=362,  substModel, substModelBF)
TKF92Pair(seq1, seq2, mu=NULL, r=NULL, distance=NULL, expectedLength=362, substModel, substModelBF)

Arguments

fasta

A named list of sequences in vector of characters format. read.fasta from package seqinr outputs this format when reading from a fasta file.

mu

A numeric value between 0 and 1 or NULL. It is the death rate per normal link in TKF92 model. When it is NULL, a joint estimation of mu, r and distance will be done. When it is given, only the distance will be estimated.

r

A numeric value between 0 and 1 or NULL. It is the success probability of the geometric distribution for modeling the fragment length in TKF92 model. When it is NULL, a joint estimation of mu, r and distance will be done. When it is given, only the distance will be estimated.

distance

A numeric value: the PAM distance between two protein sequences. When it is given, TKF92Pair only calculates the negative log-likelihood.

expectedLength

A numeric object: the expected length of input protein sequences. By default, the average sequence length, 362, from OMA browser is used.

substModel

A numeric matrix: the mutation probability from one AA to another AA at PAM distance 1. The order of AA in the matrix should be identical to AACharacterSet.

substModelBF

A vector of numeric: the backrgound frequency of AAs. The order of AA in the vector should also be identical to AACharacterSet.

seq1, seq2

A vector of character: the sequences of two proteins to compare.

Value

A list of matrices are returned: the matrix of estimated distances, the matrix of estimated distance variances, the matrix of negative log-likelihood between the sequences.

Details

Currently this implementation only supports the normal 20 AAs. Missing or Ambiguous characters are not supported.

References

Thorne, J.L., Kishino, H., and Felsenstein, J. (1992). Inching toward reality: an improved likelihood model of sequence evolution. J. Mol. Evol. 34, 3-16.

Gonnet, G.H., Cohen, M.A., and Benner, S.A. (1992). Exhaustive matching of the entire protein sequence database. Science 256, 1443-1445.

See Also

AACharacterSet, GONNET, GONNETBF

Examples

  
    ## This example is not tested due to running time > 5s
  data(GONNET)
  data(GONNETBF)
  library(seqinr)
  fasta <- read.fasta(file.path(system.file("extdata", package="TKF"),
                      "pair1.fasta"),
                      seqtype="AA", set.attributes=FALSE)
  ## 1D estimation: only distance
  TKF92(fasta, mu=0.0006137344, r=0.7016089061,
        substModel=GONNET, substModelBF=GONNETBF)
  
  ## 2D estimation: joint estimation of distance, mu and r
  TKF92(fasta, substModel=GONNET, substModelBF=GONNETBF)
  
  ## only apply to a pair of sequences
  seq1 <- fasta[[1]]
  seq2 <- fasta[[2]]
  TKF92Pair(seq1, seq2, mu=0.0006137344, r=0.7016089061,
            substModel=GONNET, substModelBF=GONNETBF)