|
|
eprotpars |
Please help by correcting and extending the Wiki pages.
This program infers an unrooted phylogeny from protein sequences, using a new method intermediate between the approaches of Eck and Dayhoff (1966) and Fitch (1971). Eck and Dayhoff (1966) allowed any amino acid to change to any other, and counted the number of such changes needed to evolve the protein sequences on each given phylogeny. This has the problem that it allows replacements which are not consistent with the genetic code, counting them equally with replacements that are consistent. Fitch, on the other hand, counted the minimum number of nucleotide substitutions that would be needed to achieve the given protein sequences. This counts silent changes equally with those that change the amino acid.
The present method insists that any changes of amino acid be consistent with the genetic code so that, for example, lysine is allowed to change to methionine but not to proline. However, changes between two amino acids via a third are allowed and counted as two changes if each of the two replacements is individually allowed. This sometimes allows changes that at first sight you would think should be outlawed. Thus we can change from phenylalanine to glutamine via leucine in two steps total. Consulting the genetic code, you will find that there is a leucine codon one step away from a phenylalanine codon, and a leucine codon one step away from glutamine. But they are not the same leucine codon. It actually takes three base substitutions to get from either of the phenylalanine codons UUU and UUC to either of the glutamine codons CAA or CAG. Why then does this program count only two? The answer is that recent DNA sequence comparisons seem to show that synonymous changes are considerably faster and easier than ones that change the amino acid. We are assuming that, in effect, synonymous changes occur so much more readily that they need not be counted. Thus, in the chain of changes UUU (Phe) --> CUU (Leu) --> CUA (Leu) --> CAA (Glu), the middle one is not counted because it does not change the amino acid (leucine).
To maintain consistency with the genetic code, it is necessary for the program internally to treat serine as two separate states (ser1 and ser2) since the two groups of serine codons are not adjacent in the code. Changes to the state "deletion" are counted as three steps to prevent the algorithm from assuming unnecessary deletions. The state "unknown" is simply taken to mean that the amino acid, which has not been determined, will in each tree that is evaluated be assumed be whichever one causes the fewest steps.
The assumptions of this method (which has not been described in the literature), are thus something like this:
1. Change in different sites is independent.
2. Change in different lineages is independent.
3. The probability of a base substitution that changes the amino acid
sequence is small over the lengths of time involved in a branch of
the phylogeny.
4. The expected amounts of change in different branches of the
phylogeny do not vary by so much that two changes in a high-rate
branch are more probable than one change in a low-rate branch.
5. The expected amounts of change do not vary enough among sites that
two changes in one site are more probable than one change in another.
6. The probability of a base change that is synonymous is much higher
than the probability of a change that is not synonymous.
That these are the assumptions of parsimony methods has been documented in a
series of papers of mine: (1973a, 1978b, 1979, 1981b, 1983b, 1988b). For an
opposing view arguing that the parsimony methods make no substantive
assumptions such as these, see the works by Farris (1983) and Sober (1983a,
1983b, 1988), but also read the exchange between Felsenstein and Sober (1986).The protein sequences are given by the one-letter code used by described in the Molecular Sequence Programs documentation file. Note that if two polypeptide chains are being used that are of different length owing to one terminating before the other, they should be coded as (say)
HIINMA*????
HIPNMGVWABT
since after the stop codon we do not definitely know that there has been a
deletion, and do not know what amino acid would have been there. If DNA
studies tell us that there is DNA sequence in that region, then we could use
"X" rather than "?". Note that "X" means an unknown amino acid, but definitely
an amino acid, while "?" could mean either that or a deletion. The distinction
is often significant in regions where there are deletions: one may want to
encode a six-base deletion as "-?????" since that way the program will only
count one deletion, not six deletion events, when the deletion arises.
However, if there are overlapping deletions it may not be so easy to know what
coding is correct.One will usually want to use "?" after a stop codon, if one does not know what amino acid is there. If the DNA sequence has been observed there, one probably ought to resist putting in the amino acids that this DNA would code for, and one should use "X" instead, because under the assumptions implicit in this parsimony method, changes to any noncoding sequence are much easier than changes in a coding region that change the amino acid, so that they shouldn't be counted anyway!
5 10
Alpha ABCDEFGHIK
Beta AB--EFGHIK
Gamma ?BCDSFG.??
Delta CIKDEFGHIK
Epsilon DIKDEFGHIK
% eprotpars
eprotpars
This application was modified for inclusion in EMBOSS by Ian Longden (il@sanger.ac.uk) Informatics Division, The Sanger Centre, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SA, UK.