Received by e-mail:
Quote:


HyPhy can be told to interpet character data using a custom alphabet by specifying the details in an alignment file. FASTA and PHYLIP formats do not directly provide a mechanism for defining a new character set, but HyPhy has custom instructions for this.

The little fragment below shows an alignment of 3 sequences on three characters (A,B and C) and a gap character:

Code:

$BASESET:"ABC"
$TOKEN:"-"="ABC"

>seq1
AABBCC
>seq2
A-BBAC
>seq3
CABBCC
 

The BASESET command tells hyphy to use A,B and C as character states, and the TOKEN command maps '-' to a 3-way ambiguity (or a gap).

The NEXUS format includes mechanisms to define a custom alphabet directly, e.g.

Code:

#NEXUS

BEGIN TAXA;
	DIMENSIONS NTAX = 3;
	TAXLABELS
		'seq1' 'seq2' 'seq3' ;
END;

BEGIN CHARACTERS;
	DIMENSIONS NCHAR = 6;
	FORMAT

	SYMBOLS = "A B C"

	EQUATE="- = ABC"
		MISSING=-
	;

MATRIX
	'seq1'  AABBCC
	'seq2'  A-BBAC
	'seq3'  CABBCC;
END;
 

Datafiles like this can be read and filtered by HyPhy and if one defines a 3 state rate matrix for the above example, e.g.

rateMx = {{*,a,b}
               {a,*,c}
               {b,c,*}},

(here an i,j entry is the rate of substituting character i with chartacter j, i,j=0..2. In the example above 'A' has code 0 since it was listed first, 'B' has code 1 and 'C' has code 2), then a model can be fitted by ML.

HTH,
Sergei