miércoles, mayo 06, 2009

TNT macros: Intro


One of the most powerful utilities of TNT is its macro's language. This language allows access to TNT internal variables, and tree and data manipulation. This coupled with the computer power of TNT makes a terrific combination.

Unfortunatelly, maybe because of a lack of an extensive manual, or because there are few papers that explicitly use TNT macros (but they are growing!), then this capability is ignored for several users. In this series of posts, I want to give an introduction of the many possibilities allowed by the usage of macros.

Part of this idea born after I see the book "phylogenetics with R" [1], and the constant exchange with Santi Catalano about the TNT's wiki. I hope to post part of this series on the TNT's wiki, but for the moment, just to get the writing feeling, I post it first on my blog ;).

The style is somewhat orientated to programming--I'm powerfully influenced by Kenighan and Ritchie :)--because I think it is the best way to learn the language.

Notation

The TNT macro language is an interpreted language, this is, each instruction is executed as it is readed, then it is possible to “write” the programs in real time (just like the old BASIC, lisp, or the fashionable python).

This can be very nice, for example, to make some simple mathematical operations directly on TNT.

An example:
> macro = ;
Macro language is ON
> var: dest ;
> set dest 4 + 5 ;
> quote 'dest' ;
9
This simple calculator can introduce us in some particularities of the language.

To activate macros, the command macro = ; must be invoked, macro - ; deactivates the macros.

Each instruction finish with a semi colon (;). Although not necessary, for clarity, an space before the semicolon is a good practice.

The keyword var is used to declare variables.

The keyword set assigns a value to a variable. In the example it is a sum. Set accepts the four basic operations, as well as more complex combinations using parenthesis.

The command quote prints on the screen. In TNT to access the value stored into a variable, the name of the variable must be between simple quotes ('). Note that set assigns a value, so the name is written without quotations, but if the value assigned is stored into a variable, then that values must be in quotations:
set dest 'first' * 'second' ;
This operation assigns to dest the value of first times second.

But more important, is that TNT macros can be written on separate file, and executing like any commando of the program.

If the macro is saved with the ".run" extention, and is on the same working directory of the program (for example c:\bin\tnt) it is possible to call the macro just typing the name of the file. For example, TNT is distributed with the macro stats.run, that calculates consistency and retention index of the trees in memory. To execute it, only is necessary to type:
> stats ;
in the program command line, and the macro runs automatically.

As any file of TNT it can be accessed using proc command (but you lost the parameter list) of the macro. It is better to use the command run (or just write the macro's name). As any command macros can be accessed trough the OS command line.

Exercises

As exercises try to use the different ways to call a macro, as well as accessing values with the commands, think on scripts like this:
macro = ;
var:
numIts
numTrees
itTrees;

set numIts 100 ;
set itTrees 10 ;
set numTrees 'numIts' * 'itTrees' ;

rseed 0;
hold 'numTrees' ;
mult = replic 'numIts' hold 'itTrees' ;
nel * ;
p/ ;
That does noting that you can do more easily by hand, but allows you to get familiarity with the notation.

References
[1] Paradis, E. 2006. Analysis of phylogenetics and evolution with R. Springer

No hay comentarios.: