outfile [slr.res]: File to which results are written. If the file already exists, it will be overwritten.
reoptimise [1]: Should the branch lengths, omega and kappa be reoptimized?
kappa [2.0]: Value for kappa. If 'reoptimise' is specified, the value given will be used as am initial estimate.
omega [0.1]: Value for omega (dN/dS). If 'reoptimise' is specified, the value given will be used as an initial estimate.
codonf [0]: How codon frequencies are estimated:
0: F61/F60 Estimates used are the empirical frequencies from the data.
1: F3x4 The frequencies of nucleotides at each codon position are estimated from the data and then multiplied together to get the frequency of observing a given codon. The frequency of stop codons is set to zero, and all other frequencies scaled appropriately.
2: F1x4 Nucleotide frequencies are estimated from the data (not taking into account at which position in the codon it occurs). The nucleotide frequencies are multiplied together to get the frequency of observing and then corrected for stop codons.
freqtype [0]: How codon frequencies are incorporated into the substitution matrix.
0: q_{ij} = pi_{j} s_{ij} → tradition method of incorporating equilibrium frequencies into subsitution matrices (Felsenstein 1981; Goldman and Yang, 1994)
1: q_{ij} = \sqrt(pi_j/pi_i) s_{ij} → described by Goldman and Whelan (2002), in this case with the additional parameter set to 0.5.
2: q_{ij} = \pi_{n} s_{ij}, where n is the nucleotide that the subsitution is to. → suggested by Muse and Gaut (1994).
3: q_{ij} = s_{ij} / pi_i. → included as an experiment, originally suggested by Bret Larget. It does not appear to describe evolution very successfully and should not be used for analyses. Kosakovsky-Pond has repeatedly stated that he finds incorporating codon frequencies in the manner of option 2 to be superior to option 0. We find that option 1 tends to perform better than either of these options.
gencode [universal]: Which genetic code to use when determining whether a given mutation is synonymous or nonsynonymous. Currently only “universal” and “mammalian” mitochondrial are supported.
nucleof [0]: Allow for empirical exchangabilities for nucleotide substitution.
0: No adjustment. All nucleotides treated the same, modulo transition / transversion.
1: The rate at which a substitution caused a mutation from nucleotide a to nucleotide b is adjust by a constant N_{ab}. This adjustment is in addition to other adjustments (e.g. transition / transversion or base frequencies).
aminof [0]: Incorporate amino acid similarity parameters into substitution matrix, adjusting omega for a change between amino acid i and amino acid j. A_{ij} is a symmetric matrix of constants representing amino acid similarities.
0: Constant omega for all amino acid changes
1: omega_{ij} = omega^{A_{ij}} → has the same form as the original codon subsitution model proposed by Goldman and Yang (but with potentially different constants).
2: omega_{ij} = a_{ij} log(omega) / [ 1 - exp(-a_{ij} log(omega)) ] → has a more population genetic derivtion, with omega being interpreted as the ratio of fixation probabilities.
nucfile [nuc.dat]: If nucleof is non-zero, read nucleotide substitution constants from nucfile. If this file does not exist, hard coded constants are used.
aminofile [amino.dat]: If aminof is non-zero, read amino acid similarity constants from aminofile. If this file does not exist, hard coded constants are used.
timemem [0]: Print summary of real time and CPU time used. Will eventually print summary of memory use as well.
ldiff [3.841459]: Twice log-likelihood difference used as a threshold for calculating support (confidence) intervals for sitewise omega estimates. This value should be the quantile from a chi-square distribution with one degree of freedom corresponding to the support required. E.g. qchisq(0.95,1) = 3.841459
random number generator is initialised using the clock.