package Text::Tradition::Stemma;
+use Bio::Phylo::IO;
+use Encode qw( decode_utf8 );
use File::chdir;
use File::Temp;
-use IPC::Run qw/ run /;
+use Graph;
+use Graph::Convert;
+use Graph::Reader::Dot;
+use IPC::Run qw/ run binary /;
use Moose;
-use Text::Tradition::Collation::Position;
+use Text::Balanced qw/ extract_bracketed /;
has collation => (
is => 'ro',
isa => 'Text::Tradition::Collation',
required => 1,
+ weak_ref => 1,
);
-has character_matrix => (
+has graph => (
+ is => 'rw',
+ isa => 'Graph',
+ predicate => 'has_graph',
+ );
+
+has distance_trees => (
is => 'ro',
- isa => 'ArrayRef[ArrayRef[Str]]',
- writer => '_save_character_matrix',
- predicate => 'has_character_matrix',
+ isa => 'ArrayRef[Graph]',
+ writer => '_save_distance_trees',
+ predicate => 'has_distance_trees',
);
+
+sub BUILD {
+ my( $self, $args ) = @_;
+ # If we have been handed a dotfile, initialize it into a graph.
+ if( exists $args->{'dot'} ) {
+ $self->graph_from_dot( $args->{'dot'} );
+ }
+}
+sub graph_from_dot {
+ my( $self, $dotfh ) = @_;
+ # Assume utf-8
+ binmode( $dotfh, ':utf8' );
+ my $reader = Graph::Reader::Dot->new();
+ my $graph = $reader->read_graph( $dotfh );
+ $graph
+ ? $self->graph( $graph )
+ : warn "Failed to parse dot in $dotfh";
+}
+
+sub as_dot {
+ my( $self, $opts ) = @_;
+ # TODO add options for display, someday
+ # TODO see what happens with Graph::Writer::Dot someday
+ my $dgraph = Graph::Convert->as_graph_easy( $self->graph );
+ # Set some class display attributes for 'hypothetical' and 'extant' nodes
+ $dgraph->set_attribute( 'flow', 'south' );
+ foreach my $n ( $dgraph->nodes ) {
+ if( $n->attribute( 'class' ) eq 'hypothetical' ) {
+ $n->set_attribute( 'shape', 'point' );
+ $n->set_attribute( 'pointshape', 'diamond' );
+ } else {
+ $n->set_attribute( 'shape', 'ellipse' );
+ }
+ }
+
+ # Render to svg via graphviz
+ my @lines = split( /\n/, $dgraph->as_graphviz() );
+ # Add the size attribute
+ if( $opts->{'size'} ) {
+ my $sizeline = " graph [ size=\"" . $opts->{'size'} . "\" ]";
+ splice( @lines, 1, 0, $sizeline );
+ }
+ return join( "\n", @lines );
+}
+
+
+# Render the stemma as SVG.
+sub as_svg {
+ my( $self, $opts ) = @_;
+ my $dot = $self->as_dot( $opts );
+ my @cmd = qw/dot -Tsvg/;
+ my( $svg, $err );
+ my $dotfile = File::Temp->new();
+ ## TODO REMOVE
+ # $dotfile->unlink_on_destroy(0);
+ binmode $dotfile, ':utf8';
+ print $dotfile $dot;
+ push( @cmd, $dotfile->filename );
+ run( \@cmd, ">", binary(), \$svg );
+ $svg = decode_utf8( $svg );
+ return $svg;
+}
+
+sub witnesses {
+ my $self = shift;
+ my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' }
+ $self->graph->vertices;
+ return @wits;
+}
+
+#### Methods for calculating phylogenetic trees ####
+
+before 'distance_trees' => sub {
+ my $self = shift;
+ my %args = @_;
+ # TODO allow specification of method for calculating distance tree
+ if( $args{'recalc'} || !$self->has_distance_trees ) {
+ # We need to make a tree before we can return it.
+ my( $ok, $result ) = $self->run_phylip_pars();
+ if( $ok ) {
+ # Save the resulting trees
+ my $trees = _parse_newick( $result );
+ $self->_save_distance_trees( $trees );
+ } else {
+ warn "Failed to calculate distance trees: $result";
+ }
+ }
+};
+
sub make_character_matrix {
my $self = shift;
unless( $self->collation->linear ) {
my $table = $self->collation->make_alignment_table;
# Push the names of the witnesses to initialize the rows of the matrix.
my @matrix = map { [ $self->_normalize_ac( $_ ) ] } @{$table->[0]};
- $DB::single = 1;
foreach my $token_index ( 1 .. $#{$table} ) {
# First implementation: make dumb alignment table, caring about
# nothing except which reading is in which position.
push( @{$matrix[$idx]}, $chars[$idx] );
}
}
- $self->_save_character_matrix( \@matrix );
+ return \@matrix;
}
sub _normalize_ac {
my %unique = ( '__UNDEF__' => 'X',
'#LACUNA#' => '?',
);
+ my %count;
my $ctr = 0;
foreach my $word ( @$row ) {
if( $word && !exists $unique{$word} ) {
$unique{$word} = chr( 65 + $ctr );
$ctr++;
}
+ $count{$word}++ if $word;
}
+ # Try to keep variants under 8 by lacunizing any singletons.
if( scalar( keys %unique ) > 8 ) {
- warn "Have more than 8 variants on this location; pars will break";
+ foreach my $word ( keys %count ) {
+ if( $count{$word} == 1 ) {
+ $unique{$word} = '?';
+ }
+ }
+ }
+ my %u = reverse %unique;
+ if( scalar( keys %u ) > 8 ) {
+ warn "Have more than 8 variants on this location; phylip will break";
}
my @chars = map { $_ ? $unique{$_} : $unique{'__UNDEF__' } } @$row;
return @chars;
}
-sub pars_input {
+sub phylip_pars_input {
my $self = shift;
- $self->make_character_matrix unless $self->has_character_matrix;
- my $matrix = '';
- my $rows = scalar @{$self->character_matrix};
- my $columns = scalar @{$self->character_matrix->[0]} - 1;
- $matrix .= "\t$rows\t$columns\n";
- foreach my $row ( @{$self->character_matrix} ) {
- $matrix .= join( '', @$row ) . "\n";
- }
- return $matrix;
+ my $character_matrix = $self->make_character_matrix;
+ my $input = '';
+ my $rows = scalar @{$character_matrix};
+ my $columns = scalar @{$character_matrix->[0]} - 1;
+ $input .= "\t$rows\t$columns\n";
+ foreach my $row ( @{$character_matrix} ) {
+ $input .= join( '', @$row ) . "\n";
+ }
+ return $input;
}
-sub run_pars {
+sub run_phylip_pars {
my $self = shift;
# Set up a temporary directory for all the default Phylip files.
my $phylip_dir = File::Temp->newdir();
- print STDERR $phylip_dir . "\n";
# $phylip_dir->unlink_on_destroy(0);
# We need an infile, and we need a command input file.
open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile";
- print MATRIX $self->pars_input();
+ print MATRIX $self->phylip_pars_input();
close MATRIX;
open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile";
return( undef, join( '', @error ) );
}
+sub _parse_newick {
+ my $newick = shift;
+ my @trees;
+ # Parse the result into a tree
+ my $forest = Bio::Phylo::IO->parse(
+ -format => 'newick',
+ -string => $newick,
+ );
+ # Turn the tree into a graph, starting with the root node
+ foreach my $tree ( @{$forest->get_entities} ) {
+ push( @trees, _graph_from_bio( $tree ) );
+ }
+ return \@trees;
+}
+
+sub _graph_from_bio {
+ my $tree = shift;
+ my $graph = Graph->new( 'undirected' => 1 );
+ # Give all the intermediate anonymous nodes a name.
+ my $i = 0;
+ foreach my $n ( @{$tree->get_entities} ) {
+ next if $n->get_name;
+ $n->set_name( $i++ );
+ }
+ my $root = $tree->get_root->get_name;
+ $graph->add_vertex( $root );
+ _add_tree_children( $graph, $root, $tree->get_root->get_children() );
+ return $graph;
+}
+
+sub _add_tree_children {
+ my( $graph, $parent, $tree_children ) = @_;
+ foreach my $c ( @$tree_children ) {
+ my $child = $c->get_name;
+ $graph->add_vertex( $child );
+ $graph->add_path( $parent, $child );
+ _add_tree_children( $graph, $child, $c->get_children() );
+ }
+}
+
no Moose;
__PACKAGE__->meta->make_immutable;
projects / scpubgit/stemmatology.git / blobdiff