Commit | Line | Data |
9463b0bf |
1 | package Text::Tradition::Stemma; |
2 | |
3 | use File::chdir; |
4 | use File::Temp; |
5 | use IPC::Run qw/ run /; |
6 | use Moose; |
7 | use Text::Tradition::Collation::Position; |
e05997e2 |
8 | use Graph; |
9 | use Graph::Reader::Dot; |
9463b0bf |
10 | |
11 | has collation => ( |
12 | is => 'ro', |
13 | isa => 'Text::Tradition::Collation', |
14 | required => 1, |
15 | ); |
16 | |
17 | has character_matrix => ( |
18 | is => 'ro', |
19 | isa => 'ArrayRef[ArrayRef[Str]]', |
20 | writer => '_save_character_matrix', |
21 | predicate => 'has_character_matrix', |
22 | ); |
e05997e2 |
23 | |
24 | has graph => ( |
25 | is => 'rw', |
26 | isa => 'Graph', |
27 | predicate => 'has_graph', |
28 | ); |
29 | |
30 | has apsp => ( |
31 | is => 'rw', |
32 | isa => 'Graph', |
33 | ); |
34 | |
35 | sub BUILD { |
36 | my( $self, $args ) = @_; |
37 | # If we have been handed a dotfile, initialize it into a graph. |
38 | if( exists $args->{'dot'} ) { |
39 | my $reader = Graph::Reader::Dot->new(); |
40 | my $graph = $reader->read_graph( $args->{'dot'} ); |
41 | $graph |
42 | ? $self->graph( $graph ) |
43 | : warn "Failed to parse dot file " . $args->{'dot'}; |
44 | } |
45 | |
46 | # If we have a graph, calculate all the shortest paths between nodes, |
47 | # disregarding direction. |
48 | if( $self->has_graph ) { |
49 | my $undirected; |
50 | if( $self->graph->is_directed ) { |
51 | # Make an undirected version. |
52 | $undirected = Graph->new( 'undirected' => 1 ); |
53 | foreach my $v ( $self->graph->vertices ) { |
54 | $undirected->add_vertex( $v ); |
55 | } |
56 | foreach my $e ( $self->graph->edges ) { |
57 | $undirected->add_edge( @$e ); |
58 | } |
59 | } else { |
60 | $undirected = $self->graph; |
61 | } |
62 | $self->apsp( $undirected->APSP_Floyd_Warshall() ); |
63 | } |
64 | } |
65 | |
66 | |
9463b0bf |
67 | sub make_character_matrix { |
68 | my $self = shift; |
69 | unless( $self->collation->linear ) { |
910a0a6d |
70 | warn "Need a linear graph in order to make an alignment table"; |
71 | return; |
9463b0bf |
72 | } |
910a0a6d |
73 | my $table = $self->collation->make_alignment_table; |
74 | # Push the names of the witnesses to initialize the rows of the matrix. |
75 | my @matrix = map { [ $self->_normalize_ac( $_ ) ] } @{$table->[0]}; |
76 | $DB::single = 1; |
77 | foreach my $token_index ( 1 .. $#{$table} ) { |
78 | # First implementation: make dumb alignment table, caring about |
79 | # nothing except which reading is in which position. |
80 | my @chars = convert_characters( $table->[$token_index] ); |
81 | foreach my $idx ( 0 .. $#matrix ) { |
82 | push( @{$matrix[$idx]}, $chars[$idx] ); |
83 | } |
9463b0bf |
84 | } |
910a0a6d |
85 | $self->_save_character_matrix( \@matrix ); |
86 | } |
9463b0bf |
87 | |
910a0a6d |
88 | sub _normalize_ac { |
89 | my( $self, $witname ) = @_; |
90 | my $ac = $self->collation->ac_label; |
91 | if( $witname =~ /(.*)\Q$ac\E$/ ) { |
92 | $witname = $1 . '_ac'; |
9463b0bf |
93 | } |
910a0a6d |
94 | return sprintf( "%-10s", $witname ); |
9463b0bf |
95 | } |
9463b0bf |
96 | |
910a0a6d |
97 | sub convert_characters { |
98 | my $row = shift; |
9463b0bf |
99 | # This is a simple algorithm that treats every reading as different. |
100 | # Eventually we will want to be able to specify how relationships |
101 | # affect the character matrix. |
eca16057 |
102 | my %unique = ( '__UNDEF__' => 'X', |
103 | '#LACUNA#' => '?', |
104 | ); |
910a0a6d |
105 | my $ctr = 0; |
106 | foreach my $word ( @$row ) { |
107 | if( $word && !exists $unique{$word} ) { |
108 | $unique{$word} = chr( 65 + $ctr ); |
109 | $ctr++; |
110 | } |
9463b0bf |
111 | } |
910a0a6d |
112 | if( scalar( keys %unique ) > 8 ) { |
113 | warn "Have more than 8 variants on this location; pars will break"; |
114 | } |
115 | my @chars = map { $_ ? $unique{$_} : $unique{'__UNDEF__' } } @$row; |
116 | return @chars; |
9463b0bf |
117 | } |
118 | |
f6066bac |
119 | sub pars_input { |
9463b0bf |
120 | my $self = shift; |
121 | $self->make_character_matrix unless $self->has_character_matrix; |
f6066bac |
122 | my $matrix = ''; |
123 | my $rows = scalar @{$self->character_matrix}; |
124 | my $columns = scalar @{$self->character_matrix->[0]} - 1; |
125 | $matrix .= "\t$rows\t$columns\n"; |
126 | foreach my $row ( @{$self->character_matrix} ) { |
910a0a6d |
127 | $matrix .= join( '', @$row ) . "\n"; |
f6066bac |
128 | } |
129 | return $matrix; |
130 | } |
131 | |
132 | sub run_pars { |
133 | my $self = shift; |
9463b0bf |
134 | |
135 | # Set up a temporary directory for all the default Phylip files. |
136 | my $phylip_dir = File::Temp->newdir(); |
eca16057 |
137 | print STDERR $phylip_dir . "\n"; |
138 | # $phylip_dir->unlink_on_destroy(0); |
f6066bac |
139 | # We need an infile, and we need a command input file. |
9463b0bf |
140 | open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile"; |
f6066bac |
141 | print MATRIX $self->pars_input(); |
9463b0bf |
142 | close MATRIX; |
143 | |
144 | open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile"; |
145 | ## TODO any configuration parameters we want to set here |
146 | # U Search for best tree? Yes |
147 | # S Search option? More thorough search |
148 | # V Number of trees to save? 100 |
149 | # J Randomize input order of species? No. Use input order |
150 | # O Outgroup root? No, use as outgroup species 1 |
151 | # T Use Threshold parsimony? No, use ordinary parsimony |
152 | # W Sites weighted? No |
153 | # M Analyze multiple data sets? No |
154 | # I Input species interleaved? Yes |
155 | # 0 Terminal type (IBM PC, ANSI, none)? ANSI |
156 | # 1 Print out the data at start of run No |
157 | # 2 Print indications of progress of run Yes |
158 | # 3 Print out tree Yes |
159 | # 4 Print out steps in each site No |
160 | # 5 Print character at all nodes of tree No |
161 | # 6 Write out trees onto tree file? Yes |
162 | print CMD "Y\n"; |
163 | close CMD; |
164 | |
165 | # And then we run the program. |
166 | ### HACKY HACKY |
167 | my $PHYLIP_PATH = '/Users/tla/Projects/phylip-3.69/exe'; |
168 | my $program = "pars"; |
169 | if( $^O eq 'darwin' ) { |
910a0a6d |
170 | $program = "$PHYLIP_PATH/$program.app/Contents/MacOS/$program"; |
9463b0bf |
171 | } else { |
910a0a6d |
172 | $program = "$PHYLIP_PATH/$program"; |
9463b0bf |
173 | } |
174 | |
175 | { |
910a0a6d |
176 | # We need to run it in our temporary directory where we have created |
177 | # all the expected files. |
178 | local $CWD = $phylip_dir; |
179 | my @cmd = ( $program ); |
180 | run \@cmd, '<', 'cmdfile', '>', '/dev/null'; |
9463b0bf |
181 | } |
182 | # Now our output should be in 'outfile' and our tree in 'outtree', |
183 | # both in the temp directory. |
184 | |
185 | my @outtree; |
186 | if( -f "$phylip_dir/outtree" ) { |
910a0a6d |
187 | open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read"; |
188 | @outtree = <TREE>; |
189 | close TREE; |
9463b0bf |
190 | } |
191 | return( 1, join( '', @outtree ) ) if @outtree; |
192 | |
193 | my @error; |
f6066bac |
194 | if( -f "$phylip_dir/outfile" ) { |
910a0a6d |
195 | open( OUTPUT, "$phylip_dir/outfile" ) or die "Could not open output for read"; |
196 | @error = <OUTPUT>; |
197 | close OUTPUT; |
9463b0bf |
198 | } else { |
910a0a6d |
199 | push( @error, "Neither outtree nor output file was produced!" ); |
9463b0bf |
200 | } |
201 | return( undef, join( '', @error ) ); |
202 | } |
203 | |
204 | no Moose; |
205 | __PACKAGE__->meta->make_immutable; |
206 | |
207 | 1; |