1 package Text::Tradition::Stemma;
4 use Encode qw( decode_utf8 );
9 use Graph::Reader::Dot;
10 use IPC::Run qw/ run binary /;
12 use Text::Balanced qw/ extract_bracketed /;
16 isa => 'Text::Tradition::Collation',
23 predicate => 'has_graph',
26 has distance_trees => (
28 isa => 'ArrayRef[Graph]',
29 writer => '_save_distance_trees',
30 predicate => 'has_distance_trees',
34 my( $self, $args ) = @_;
35 # If we have been handed a dotfile, initialize it into a graph.
36 if( exists $args->{'dot'} ) {
37 # Open the file, assume UTF-8
38 open( my $dot, $args->{'dot'} ) or warn "Failed to read dot file";
39 # TODO don't bother if we haven't opened
40 binmode $dot, ":utf8";
41 my $reader = Graph::Reader::Dot->new();
42 my $graph = $reader->read_graph( $dot );
44 ? $self->graph( $graph )
45 : warn "Failed to parse dot file " . $args->{'dot'};
49 # Render the stemma as SVG.
51 my( $self, $opts ) = @_;
52 # TODO add options for display, someday
53 my $dgraph = Graph::Convert->as_graph_easy( $self->graph );
54 # Set some class display attributes for 'hypothetical' and 'extant' nodes
55 $dgraph->set_attribute( 'flow', 'south' );
56 foreach my $n ( $dgraph->nodes ) {
57 if( $n->attribute( 'class' ) eq 'hypothetical' ) {
58 $n->set_attribute( 'shape', 'point' );
59 $n->set_attribute( 'pointshape', 'diamond' );
61 $n->set_attribute( 'shape', 'ellipse' );
65 # Render to svg via graphviz
66 my @lines = split( /\n/, $dgraph->as_graphviz() );
67 # Add the size attribute
68 if( $opts->{'size'} ) {
69 my $sizeline = " graph [ size=\"" . $opts->{'size'} . "\" ]";
70 splice( @lines, 1, 0, $sizeline );
72 my @cmd = qw/dot -Tsvg/;
74 my $dotfile = File::Temp->new();
76 # $dotfile->unlink_on_destroy(0);
77 binmode $dotfile, ':utf8';
78 print $dotfile join( "\n", @lines );
79 push( @cmd, $dotfile->filename );
80 run( \@cmd, ">", binary(), \$svg );
81 $svg = decode_utf8( $svg );
87 my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' }
88 $self->graph->vertices;
92 #### Methods for calculating phylogenetic trees ####
94 before 'distance_trees' => sub {
97 # TODO allow specification of method for calculating distance tree
98 if( $args{'recalc'} || !$self->has_distance_trees ) {
99 # We need to make a tree before we can return it.
100 my( $ok, $result ) = $self->run_phylip_pars();
102 # Save the resulting trees
103 my $trees = _parse_newick( $result );
104 $self->_save_distance_trees( $trees );
106 warn "Failed to calculate distance trees: $result";
111 sub make_character_matrix {
113 unless( $self->collation->linear ) {
114 warn "Need a linear graph in order to make an alignment table";
117 my $table = $self->collation->make_alignment_table;
118 # Push the names of the witnesses to initialize the rows of the matrix.
119 my @matrix = map { [ $self->_normalize_ac( $_ ) ] } @{$table->[0]};
120 foreach my $token_index ( 1 .. $#{$table} ) {
121 # First implementation: make dumb alignment table, caring about
122 # nothing except which reading is in which position.
123 my @chars = convert_characters( $table->[$token_index] );
124 foreach my $idx ( 0 .. $#matrix ) {
125 push( @{$matrix[$idx]}, $chars[$idx] );
132 my( $self, $witname ) = @_;
133 my $ac = $self->collation->ac_label;
134 if( $witname =~ /(.*)\Q$ac\E$/ ) {
135 $witname = $1 . '_ac';
137 return sprintf( "%-10s", $witname );
140 sub convert_characters {
142 # This is a simple algorithm that treats every reading as different.
143 # Eventually we will want to be able to specify how relationships
144 # affect the character matrix.
145 my %unique = ( '__UNDEF__' => 'X',
150 foreach my $word ( @$row ) {
151 if( $word && !exists $unique{$word} ) {
152 $unique{$word} = chr( 65 + $ctr );
155 $count{$word}++ if $word;
157 # Try to keep variants under 8 by lacunizing any singletons.
158 if( scalar( keys %unique ) > 8 ) {
159 foreach my $word ( keys %count ) {
160 if( $count{$word} == 1 ) {
161 $unique{$word} = '?';
165 my %u = reverse %unique;
166 if( scalar( keys %u ) > 8 ) {
167 warn "Have more than 8 variants on this location; phylip will break";
169 my @chars = map { $_ ? $unique{$_} : $unique{'__UNDEF__' } } @$row;
173 sub phylip_pars_input {
175 my $character_matrix = $self->make_character_matrix;
177 my $rows = scalar @{$character_matrix};
178 my $columns = scalar @{$character_matrix->[0]} - 1;
179 $input .= "\t$rows\t$columns\n";
180 foreach my $row ( @{$character_matrix} ) {
181 $input .= join( '', @$row ) . "\n";
186 sub run_phylip_pars {
189 # Set up a temporary directory for all the default Phylip files.
190 my $phylip_dir = File::Temp->newdir();
191 # $phylip_dir->unlink_on_destroy(0);
192 # We need an infile, and we need a command input file.
193 open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile";
194 print MATRIX $self->phylip_pars_input();
197 open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile";
198 ## TODO any configuration parameters we want to set here
199 # U Search for best tree? Yes
200 # S Search option? More thorough search
201 # V Number of trees to save? 100
202 # J Randomize input order of species? No. Use input order
203 # O Outgroup root? No, use as outgroup species 1
204 # T Use Threshold parsimony? No, use ordinary parsimony
205 # W Sites weighted? No
206 # M Analyze multiple data sets? No
207 # I Input species interleaved? Yes
208 # 0 Terminal type (IBM PC, ANSI, none)? ANSI
209 # 1 Print out the data at start of run No
210 # 2 Print indications of progress of run Yes
211 # 3 Print out tree Yes
212 # 4 Print out steps in each site No
213 # 5 Print character at all nodes of tree No
214 # 6 Write out trees onto tree file? Yes
218 # And then we run the program.
220 my $PHYLIP_PATH = '/Users/tla/Projects/phylip-3.69/exe';
221 my $program = "pars";
222 if( $^O eq 'darwin' ) {
223 $program = "$PHYLIP_PATH/$program.app/Contents/MacOS/$program";
225 $program = "$PHYLIP_PATH/$program";
229 # We need to run it in our temporary directory where we have created
230 # all the expected files.
231 local $CWD = $phylip_dir;
232 my @cmd = ( $program );
233 run \@cmd, '<', 'cmdfile', '>', '/dev/null';
235 # Now our output should be in 'outfile' and our tree in 'outtree',
236 # both in the temp directory.
239 if( -f "$phylip_dir/outtree" ) {
240 open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read";
244 return( 1, join( '', @outtree ) ) if @outtree;
247 if( -f "$phylip_dir/outfile" ) {
248 open( OUTPUT, "$phylip_dir/outfile" ) or die "Could not open output for read";
252 push( @error, "Neither outtree nor output file was produced!" );
254 return( undef, join( '', @error ) );
260 # Parse the result into a tree
261 my $forest = Bio::Phylo::IO->parse(
265 # Turn the tree into a graph, starting with the root node
266 foreach my $tree ( @{$forest->get_entities} ) {
267 push( @trees, _graph_from_bio( $tree ) );
272 sub _graph_from_bio {
274 my $graph = Graph->new( 'undirected' => 1 );
275 # Give all the intermediate anonymous nodes a name.
277 foreach my $n ( @{$tree->get_entities} ) {
278 next if $n->get_name;
279 $n->set_name( $i++ );
281 my $root = $tree->get_root->get_name;
282 $graph->add_vertex( $root );
283 _add_tree_children( $graph, $root, $tree->get_root->get_children() );
287 sub _add_tree_children {
288 my( $graph, $parent, $tree_children ) = @_;
289 foreach my $c ( @$tree_children ) {
290 my $child = $c->get_name;
291 $graph->add_vertex( $child );
292 $graph->add_path( $parent, $child );
293 _add_tree_children( $graph, $child, $c->get_children() );
298 __PACKAGE__->meta->make_immutable;