DART: DNA/Amino/RNA Tests
===============================================================================

The primary reference point for DART is the following URL:

http://biowiki.org/dart

===============================================================================

INSTALLATION
============

Type "./configure; make all".

See INSTALL file for more info.

===============================================================================

NOTES
=====

DART is a collection of application programs for doing
probabilistic bioinformatics using evolutionary models
and stochastic grammars.

As well as the primary packages (Handel, Stemloc, XRate etc),
DART contains various utilities related to the primary datatypes:
DNA, RNA and protein sequence.

The package was mostly written by Ian Holmes <ihh@berkeley.edu>
For a full list of contributors see http://biowiki.org/dart

Some free packages from external sources are included in the distribution:
	dart/src/newmat          [Newmat, by RB Davies]
	dart/src/randlib         [randlib, from UTexas Biomath dept] 
	dart/src/Weighbor        [Weighbor, by Bill Bruno & Nick Socci]
	dart/src/util/Regexp.*   [regexp library by Henry Spencer & others]

DART is released under the GNU Public License (GPL).

===============================================================================

Supported packages
==================

xrate:   A package for training phylo-grammars using EM,
          and applying them to annotate alignments.

Handel:  phylogenetic multiple alignment software based on
          the TKF91 evolutionary indel/substitution model.

Stemloc: comparative RNA structure-finder using accelerated
          pairwise stochastic context-free grammars

evoldoer: pairwise RNA alignment using an evolutionary model
          (formerly "tkfstalign")

protpal: approximate phylogenetic alignment using sequence profiles
         


Also included are various bioinformatics tools, tests and algorithms.


Longer (but not by much) package descriptions
=============================================

** xrate

This is a package for training & annotation using phylogrammars.
See the following URL for documentation:

http://biowiki.org/XrateSoftware


** Handel

This is a package for doing multiple sequence alignment under an
evolutionary model.

Since it uses a probabilistic (MCMC) approach, as well as "greedy"
heuristics (progressive alignment etc),	the program can generate
suboptimal alignments as well as looking for the best alignment.

The probabilistic framework also means that the program can be
"trained" directly from data, avoiding the need to supply
"gap penalties", "substitution matrices" etc.

Currently the evolutionary model used is the TKF model.
This implies global alignment, homogenous selection & linear gap costs.
A more realistic model (affine gaps, local alignment, heterogenous)
is under development.

The main programs in this package are:


tkfemit -- generates alignments from the TKF model, given a tree
tkfdistance -- computes a PHYLIP distance matrix from unaligned sequences
tkfalign -- builds & samples alignments, given a tree
tkfidem -- EM estimation of TKF91 indel rates
phylocomposer -- phylogenetic transducer composition & alignment
phylodirector.pl -- animations of phylo-transducers
weighbor -- Bill Bruno's program to build a tree from the tkfdistance matrix



** Stemloc

This is a program for finding conserved motifs in RNA sequences,
using pairwise stochastic context-free grammars.

The algorithm simultaneously aligns and folds multiple RNA sequences,
and may be viewed as a constrained (and so accelerated) version of
the dynamic programming	method of Sankoff et al.

Again, due to the probabilistic nature of the approach,
the program can be "trained" directly from data, without the
need for expert knowledge to set the parameters.
	
The main program in this package is "stemloc" itself.
See the following URL for a tutorial:

http://biowiki.org/StemlocTutorial


** evoldoer

Like stemloc, this is an RNA alignment program, but it only does
pairwise alignments, not multiple alignments. It is however
fully based on an evolutionary model for RNA structure,
called the TKF Structure Tree (TKFST), based on the TKF91 model.
TKFST models indels of whole substructures as well as point
(base) substitutions & indels and covariant (basepair)
substitutions & indels.
	

** protpal
Program for approximate statistical alignment and reconstruction.  Uses progressive alignment with sequence profiles. 


REFERENCES
==========

For references see the following URL:

http://biowiki.org/PaperArchive