Commit | Line | Data |
9463b0bf |
1 | package Text::Tradition::Stemma; |
2 | |
40f19742 |
3 | use Bio::Phylo::IO; |
e79c23c7 |
4 | use Encode qw( decode_utf8 ); |
9463b0bf |
5 | use File::chdir; |
6 | use File::Temp; |
e05997e2 |
7 | use Graph; |
8 | use Graph::Reader::Dot; |
e79c23c7 |
9 | use IPC::Run qw/ run binary /; |
40f19742 |
10 | use Moose; |
9463b0bf |
11 | |
12 | has collation => ( |
13 | is => 'ro', |
14 | isa => 'Text::Tradition::Collation', |
15 | required => 1, |
8d9a1cd8 |
16 | weak_ref => 1, |
9463b0bf |
17 | ); |
18 | |
e05997e2 |
19 | has graph => ( |
20 | is => 'rw', |
21 | isa => 'Graph', |
22 | predicate => 'has_graph', |
23 | ); |
24 | |
40f19742 |
25 | has distance_trees => ( |
26 | is => 'ro', |
27 | isa => 'ArrayRef[Graph]', |
28 | writer => '_save_distance_trees', |
29 | predicate => 'has_distance_trees', |
30 | ); |
c0ccdb62 |
31 | |
e05997e2 |
32 | sub BUILD { |
33 | my( $self, $args ) = @_; |
34 | # If we have been handed a dotfile, initialize it into a graph. |
35 | if( exists $args->{'dot'} ) { |
8d9a1cd8 |
36 | $self->graph_from_dot( $args->{'dot'} ); |
e05997e2 |
37 | } |
c0ccdb62 |
38 | } |
39 | |
8d9a1cd8 |
40 | sub graph_from_dot { |
41 | my( $self, $dotfh ) = @_; |
42 | # Assume utf-8 |
43 | binmode( $dotfh, ':utf8' ); |
44 | my $reader = Graph::Reader::Dot->new(); |
45 | my $graph = $reader->read_graph( $dotfh ); |
7a7c249c |
46 | if( $graph ) { |
47 | $self->graph( $graph ); |
48 | # Go through the nodes and set any non-hypothetical node to extant. |
49 | foreach my $v ( $self->graph->vertices ) { |
50 | $self->graph->set_vertex_attribute( $v, 'class', 'extant' ) |
51 | unless $self->graph->has_vertex_attribute( $v, 'class' ); |
52 | } |
53 | } else { |
54 | warn "Failed to parse dot in $dotfh"; |
55 | } |
8d9a1cd8 |
56 | } |
57 | |
58 | sub as_dot { |
e367f5c0 |
59 | my( $self, $opts ) = @_; |
7a7c249c |
60 | |
61 | # Get default and specified options |
62 | my %graphopts = (); |
63 | my %nodeopts = ( |
64 | 'fontsize' => 11, |
65 | 'hshape' => 'plaintext', # Shape for the hypothetical nodes |
66 | 'htext' => '*', |
67 | 'style' => 'filled', |
68 | 'fillcolor' => 'white', |
69 | 'shape' => 'ellipse', # Shape for the extant nodes |
70 | ); |
71 | my %edgeopts = ( |
72 | 'arrowhead' => 'open', |
73 | ); |
74 | @graphopts{ keys %{$opts->{'graph'}} } = values %{$opts->{'graph'}} |
75 | if $opts->{'graph'}; |
76 | @nodeopts{ keys %{$opts->{'node'}} } = values %{$opts->{'node'}} |
77 | if $opts->{'node'}; |
78 | @edgeopts{ keys %{$opts->{'edge'}} } = values %{$opts->{'edge'}} |
79 | if $opts->{'edge'}; |
80 | |
81 | my @dotlines; |
82 | push( @dotlines, 'digraph stemma {' ); |
83 | ## Print out the global attributes |
84 | push( @dotlines, _make_dotline( 'graph', %graphopts ) ) if keys %graphopts; |
85 | push( @dotlines, _make_dotline( 'edge', %edgeopts ) ) if keys %edgeopts; |
86 | ## Delete our special attributes from the node set before continuing |
87 | my $hshape = delete $nodeopts{'hshape'}; |
88 | my $htext = delete $nodeopts{'htext'}; |
89 | push( @dotlines, _make_dotline( 'node', %nodeopts ) ) if keys %nodeopts; |
90 | |
91 | # Add each of the nodes. |
92 | foreach my $n ( $self->graph->vertices ) { |
93 | if( $self->graph->get_vertex_attribute( $n, 'class' ) eq 'hypothetical' ) { |
94 | # Apply our display settings for hypothetical nodes. |
95 | push( @dotlines, _make_dotline( $n, 'shape' => $hshape, 'label' => $htext ) ); |
e79c23c7 |
96 | } else { |
7a7c249c |
97 | # Use the default display settings. |
98 | push( @dotlines, " $n;" ); |
e79c23c7 |
99 | } |
100 | } |
7a7c249c |
101 | # Add each of our edges. |
102 | foreach my $e ( $self->graph->edges ) { |
103 | my( $from, $to ) = @$e; |
104 | push( @dotlines, " $from -> $to;" ); |
105 | } |
106 | push( @dotlines, '}' ); |
e79c23c7 |
107 | |
7a7c249c |
108 | return join( "\n", @dotlines ); |
109 | } |
110 | |
111 | |
112 | # Another version of dot output meant for graph editing, thus |
113 | # much simpler. |
114 | sub editable { |
115 | my $self = shift; |
116 | my @dotlines; |
117 | push( @dotlines, 'digraph stemma {' ); |
118 | my @real; # A cheap sort |
119 | foreach my $n ( sort $self->graph->vertices ) { |
120 | my $c = $self->graph->get_vertex_attribute( $n, 'class' ); |
121 | $c = 'extant' unless $c; |
122 | if( $c eq 'extant' ) { |
123 | push( @real, $n ); |
124 | } else { |
125 | push( @dotlines, _make_dotline( $n, 'class' => $c ) ); |
126 | } |
e367f5c0 |
127 | } |
7a7c249c |
128 | # Now do the real ones |
129 | foreach my $n ( @real ) { |
130 | push( @dotlines, _make_dotline( $n, 'class' => 'extant' ) ); |
131 | } |
132 | foreach my $e ( sort _by_vertex $self->graph->edges ) { |
133 | my( $from, $to ) = @$e; |
134 | push( @dotlines, " $from -> $to;" ); |
135 | } |
136 | push( @dotlines, '}' ); |
137 | return join( "\n", @dotlines ); |
138 | } |
139 | |
140 | sub _make_dotline { |
141 | my( $obj, %attr ) = @_; |
142 | my @pairs; |
143 | foreach my $k ( keys %attr ) { |
144 | my $v = $attr{$k}; |
145 | $v =~ s/\"/\\\"/g; |
146 | push( @pairs, "$k=\"$v\"" ); |
147 | } |
148 | return sprintf( " %s [ %s ];", $obj, join( ', ', @pairs ) ); |
8d9a1cd8 |
149 | } |
150 | |
7a7c249c |
151 | sub _by_vertex { |
152 | return $a->[0].$a->[1] cmp $b->[0].$b->[1]; |
153 | } |
8d9a1cd8 |
154 | |
155 | # Render the stemma as SVG. |
156 | sub as_svg { |
157 | my( $self, $opts ) = @_; |
158 | my $dot = $self->as_dot( $opts ); |
e79c23c7 |
159 | my @cmd = qw/dot -Tsvg/; |
160 | my( $svg, $err ); |
161 | my $dotfile = File::Temp->new(); |
162 | ## TODO REMOVE |
163 | # $dotfile->unlink_on_destroy(0); |
164 | binmode $dotfile, ':utf8'; |
8d9a1cd8 |
165 | print $dotfile $dot; |
e79c23c7 |
166 | push( @cmd, $dotfile->filename ); |
167 | run( \@cmd, ">", binary(), \$svg ); |
168 | $svg = decode_utf8( $svg ); |
169 | return $svg; |
170 | } |
171 | |
08e0fb85 |
172 | sub witnesses { |
173 | my $self = shift; |
174 | my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' } |
175 | $self->graph->vertices; |
176 | return @wits; |
177 | } |
178 | |
e79c23c7 |
179 | #### Methods for calculating phylogenetic trees #### |
180 | |
40f19742 |
181 | before 'distance_trees' => sub { |
182 | my $self = shift; |
183 | my %args = @_; |
184 | # TODO allow specification of method for calculating distance tree |
185 | if( $args{'recalc'} || !$self->has_distance_trees ) { |
186 | # We need to make a tree before we can return it. |
187 | my( $ok, $result ) = $self->run_phylip_pars(); |
188 | if( $ok ) { |
c0ccdb62 |
189 | # Save the resulting trees |
190 | my $trees = _parse_newick( $result ); |
191 | $self->_save_distance_trees( $trees ); |
40f19742 |
192 | } else { |
c0ccdb62 |
193 | warn "Failed to calculate distance trees: $result"; |
40f19742 |
194 | } |
195 | } |
196 | }; |
e05997e2 |
197 | |
9463b0bf |
198 | sub make_character_matrix { |
199 | my $self = shift; |
200 | unless( $self->collation->linear ) { |
910a0a6d |
201 | warn "Need a linear graph in order to make an alignment table"; |
202 | return; |
9463b0bf |
203 | } |
910a0a6d |
204 | my $table = $self->collation->make_alignment_table; |
205 | # Push the names of the witnesses to initialize the rows of the matrix. |
206 | my @matrix = map { [ $self->_normalize_ac( $_ ) ] } @{$table->[0]}; |
910a0a6d |
207 | foreach my $token_index ( 1 .. $#{$table} ) { |
208 | # First implementation: make dumb alignment table, caring about |
209 | # nothing except which reading is in which position. |
210 | my @chars = convert_characters( $table->[$token_index] ); |
211 | foreach my $idx ( 0 .. $#matrix ) { |
212 | push( @{$matrix[$idx]}, $chars[$idx] ); |
213 | } |
9463b0bf |
214 | } |
c0ccdb62 |
215 | return \@matrix; |
910a0a6d |
216 | } |
9463b0bf |
217 | |
910a0a6d |
218 | sub _normalize_ac { |
219 | my( $self, $witname ) = @_; |
220 | my $ac = $self->collation->ac_label; |
221 | if( $witname =~ /(.*)\Q$ac\E$/ ) { |
222 | $witname = $1 . '_ac'; |
9463b0bf |
223 | } |
910a0a6d |
224 | return sprintf( "%-10s", $witname ); |
9463b0bf |
225 | } |
9463b0bf |
226 | |
910a0a6d |
227 | sub convert_characters { |
228 | my $row = shift; |
9463b0bf |
229 | # This is a simple algorithm that treats every reading as different. |
230 | # Eventually we will want to be able to specify how relationships |
231 | # affect the character matrix. |
eca16057 |
232 | my %unique = ( '__UNDEF__' => 'X', |
233 | '#LACUNA#' => '?', |
234 | ); |
a7fb3133 |
235 | my %count; |
910a0a6d |
236 | my $ctr = 0; |
237 | foreach my $word ( @$row ) { |
238 | if( $word && !exists $unique{$word} ) { |
239 | $unique{$word} = chr( 65 + $ctr ); |
240 | $ctr++; |
241 | } |
a7fb3133 |
242 | $count{$word}++ if $word; |
9463b0bf |
243 | } |
a7fb3133 |
244 | # Try to keep variants under 8 by lacunizing any singletons. |
910a0a6d |
245 | if( scalar( keys %unique ) > 8 ) { |
a7fb3133 |
246 | foreach my $word ( keys %count ) { |
247 | if( $count{$word} == 1 ) { |
248 | $unique{$word} = '?'; |
249 | } |
250 | } |
251 | } |
252 | my %u = reverse %unique; |
253 | if( scalar( keys %u ) > 8 ) { |
40f19742 |
254 | warn "Have more than 8 variants on this location; phylip will break"; |
910a0a6d |
255 | } |
256 | my @chars = map { $_ ? $unique{$_} : $unique{'__UNDEF__' } } @$row; |
257 | return @chars; |
9463b0bf |
258 | } |
259 | |
40f19742 |
260 | sub phylip_pars_input { |
9463b0bf |
261 | my $self = shift; |
c0ccdb62 |
262 | my $character_matrix = $self->make_character_matrix; |
263 | my $input = ''; |
264 | my $rows = scalar @{$character_matrix}; |
265 | my $columns = scalar @{$character_matrix->[0]} - 1; |
266 | $input .= "\t$rows\t$columns\n"; |
267 | foreach my $row ( @{$character_matrix} ) { |
268 | $input .= join( '', @$row ) . "\n"; |
f6066bac |
269 | } |
c0ccdb62 |
270 | return $input; |
f6066bac |
271 | } |
272 | |
40f19742 |
273 | sub run_phylip_pars { |
f6066bac |
274 | my $self = shift; |
9463b0bf |
275 | |
276 | # Set up a temporary directory for all the default Phylip files. |
277 | my $phylip_dir = File::Temp->newdir(); |
eca16057 |
278 | # $phylip_dir->unlink_on_destroy(0); |
f6066bac |
279 | # We need an infile, and we need a command input file. |
9463b0bf |
280 | open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile"; |
40f19742 |
281 | print MATRIX $self->phylip_pars_input(); |
9463b0bf |
282 | close MATRIX; |
283 | |
284 | open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile"; |
285 | ## TODO any configuration parameters we want to set here |
286 | # U Search for best tree? Yes |
287 | # S Search option? More thorough search |
288 | # V Number of trees to save? 100 |
289 | # J Randomize input order of species? No. Use input order |
290 | # O Outgroup root? No, use as outgroup species 1 |
291 | # T Use Threshold parsimony? No, use ordinary parsimony |
292 | # W Sites weighted? No |
293 | # M Analyze multiple data sets? No |
294 | # I Input species interleaved? Yes |
295 | # 0 Terminal type (IBM PC, ANSI, none)? ANSI |
296 | # 1 Print out the data at start of run No |
297 | # 2 Print indications of progress of run Yes |
298 | # 3 Print out tree Yes |
299 | # 4 Print out steps in each site No |
300 | # 5 Print character at all nodes of tree No |
301 | # 6 Write out trees onto tree file? Yes |
302 | print CMD "Y\n"; |
303 | close CMD; |
304 | |
305 | # And then we run the program. |
306 | ### HACKY HACKY |
307 | my $PHYLIP_PATH = '/Users/tla/Projects/phylip-3.69/exe'; |
308 | my $program = "pars"; |
309 | if( $^O eq 'darwin' ) { |
910a0a6d |
310 | $program = "$PHYLIP_PATH/$program.app/Contents/MacOS/$program"; |
9463b0bf |
311 | } else { |
910a0a6d |
312 | $program = "$PHYLIP_PATH/$program"; |
9463b0bf |
313 | } |
314 | |
315 | { |
910a0a6d |
316 | # We need to run it in our temporary directory where we have created |
317 | # all the expected files. |
318 | local $CWD = $phylip_dir; |
319 | my @cmd = ( $program ); |
320 | run \@cmd, '<', 'cmdfile', '>', '/dev/null'; |
9463b0bf |
321 | } |
322 | # Now our output should be in 'outfile' and our tree in 'outtree', |
323 | # both in the temp directory. |
324 | |
325 | my @outtree; |
326 | if( -f "$phylip_dir/outtree" ) { |
910a0a6d |
327 | open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read"; |
328 | @outtree = <TREE>; |
329 | close TREE; |
9463b0bf |
330 | } |
331 | return( 1, join( '', @outtree ) ) if @outtree; |
332 | |
333 | my @error; |
f6066bac |
334 | if( -f "$phylip_dir/outfile" ) { |
910a0a6d |
335 | open( OUTPUT, "$phylip_dir/outfile" ) or die "Could not open output for read"; |
336 | @error = <OUTPUT>; |
337 | close OUTPUT; |
9463b0bf |
338 | } else { |
910a0a6d |
339 | push( @error, "Neither outtree nor output file was produced!" ); |
9463b0bf |
340 | } |
341 | return( undef, join( '', @error ) ); |
342 | } |
343 | |
40f19742 |
344 | sub _parse_newick { |
345 | my $newick = shift; |
346 | my @trees; |
347 | # Parse the result into a tree |
348 | my $forest = Bio::Phylo::IO->parse( |
349 | -format => 'newick', |
350 | -string => $newick, |
351 | ); |
352 | # Turn the tree into a graph, starting with the root node |
353 | foreach my $tree ( @{$forest->get_entities} ) { |
354 | push( @trees, _graph_from_bio( $tree ) ); |
355 | } |
356 | return \@trees; |
357 | } |
358 | |
359 | sub _graph_from_bio { |
360 | my $tree = shift; |
361 | my $graph = Graph->new( 'undirected' => 1 ); |
362 | # Give all the intermediate anonymous nodes a name. |
363 | my $i = 0; |
364 | foreach my $n ( @{$tree->get_entities} ) { |
365 | next if $n->get_name; |
366 | $n->set_name( $i++ ); |
367 | } |
368 | my $root = $tree->get_root->get_name; |
369 | $graph->add_vertex( $root ); |
370 | _add_tree_children( $graph, $root, $tree->get_root->get_children() ); |
371 | return $graph; |
372 | } |
373 | |
374 | sub _add_tree_children { |
375 | my( $graph, $parent, $tree_children ) = @_; |
376 | foreach my $c ( @$tree_children ) { |
377 | my $child = $c->get_name; |
378 | $graph->add_vertex( $child ); |
379 | $graph->add_path( $parent, $child ); |
380 | _add_tree_children( $graph, $child, $c->get_children() ); |
381 | } |
382 | } |
383 | |
9463b0bf |
384 | no Moose; |
385 | __PACKAGE__->meta->make_immutable; |
386 | |
387 | 1; |