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; |
8 | |
9 | has collation => ( |
10 | is => 'ro', |
11 | isa => 'Text::Tradition::Collation', |
12 | required => 1, |
13 | ); |
14 | |
15 | has character_matrix => ( |
16 | is => 'ro', |
17 | isa => 'ArrayRef[ArrayRef[Str]]', |
18 | writer => '_save_character_matrix', |
19 | predicate => 'has_character_matrix', |
20 | ); |
21 | |
22 | sub make_character_matrix { |
23 | my $self = shift; |
24 | unless( $self->collation->linear ) { |
25 | warn "Need a linear graph in order to make an alignment table"; |
26 | return; |
27 | } |
28 | my @all_pos = sort { Text::Tradition::Collation::Position::str_cmp( $a, $b ) } |
29 | $self->collation->possible_positions; |
30 | my $table = []; |
31 | my $characters = {}; |
32 | map { $characters->{$_} = {} } @all_pos; |
33 | foreach my $wit ( @{$self->collation->tradition->witnesses} ) { |
34 | # First implementation: make dumb alignment table, caring about |
35 | # nothing except which reading is in which position. |
36 | push( @$table, [ $wit->sigil, make_witness_row( $characters, $wit->path, |
37 | \@all_pos ) ] ); |
38 | if( $wit->has_ante_corr ) { |
39 | push( @$table, [ $wit->sigil . "_ac", |
40 | make_witness_row( $characters, $wit->uncorrected_path, |
41 | \@all_pos ) ] ); |
42 | } |
43 | } |
44 | $self->_save_character_matrix( $table ); |
45 | } |
46 | |
47 | sub make_witness_row { |
48 | my( $characters, $path, $positions ) = @_; |
49 | my @row; |
50 | my $pathdrift = 0; |
51 | foreach my $i( 0 .. $#{$positions} ) { |
52 | if( $path->[$i-$pathdrift]->position->minref eq $positions->[$i] ) { |
53 | push( @row, get_character( $path->[$i-$pathdrift], $characters ) ); |
54 | } else { |
55 | push( @row, 'X' ); |
56 | $pathdrift++; |
57 | } |
58 | $i++; |
59 | } |
60 | return @row; |
61 | } |
62 | |
63 | |
64 | sub get_character { |
65 | my( $reading, $characters ) = @_; |
66 | my $this_pos = $characters->{$reading->position->minref}; |
67 | # This is a simple algorithm that treats every reading as different. |
68 | # Eventually we will want to be able to specify how relationships |
69 | # affect the character matrix. |
70 | my $text = $reading->text; |
71 | unless( exists $this_pos->{$text} ) { |
72 | # We need to find what the next character is here, and record it. |
73 | my @all_chr = sort { $a <=> $b } values( %$this_pos ); |
74 | if( @all_chr == 8 ) { |
75 | warn "Already have eight variants at position " |
76 | . $reading->position->minref . "; not adding " . $reading->text; |
77 | return '?'; |
78 | } |
79 | $this_pos->{$text} = scalar @all_chr; |
80 | } |
81 | return $this_pos->{$text}; |
82 | } |
83 | |
84 | sub run_pars { |
85 | my $self = shift; |
86 | $self->make_character_matrix unless $self->has_character_matrix; |
87 | |
88 | # Set up a temporary directory for all the default Phylip files. |
89 | my $phylip_dir = File::Temp->newdir(); |
90 | |
91 | # We need an infile, and we need a command input file. |
92 | $DB::single = 1; |
93 | open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile"; |
94 | my $rows = scalar @{$self->character_matrix}; |
95 | my $columns = scalar @{$self->character_matrix->[0]} - 1; |
96 | print MATRIX "\t$rows\t$columns\n"; |
97 | foreach my $row ( @{$self->character_matrix} ) { |
98 | my $wit = shift @$row; |
99 | my $chars = join( '', @$row ); |
100 | print MATRIX sprintf( "%-10s%s\n", $wit, $chars ); |
101 | } |
102 | close MATRIX; |
103 | |
104 | open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile"; |
105 | ## TODO any configuration parameters we want to set here |
106 | # U Search for best tree? Yes |
107 | # S Search option? More thorough search |
108 | # V Number of trees to save? 100 |
109 | # J Randomize input order of species? No. Use input order |
110 | # O Outgroup root? No, use as outgroup species 1 |
111 | # T Use Threshold parsimony? No, use ordinary parsimony |
112 | # W Sites weighted? No |
113 | # M Analyze multiple data sets? No |
114 | # I Input species interleaved? Yes |
115 | # 0 Terminal type (IBM PC, ANSI, none)? ANSI |
116 | # 1 Print out the data at start of run No |
117 | # 2 Print indications of progress of run Yes |
118 | # 3 Print out tree Yes |
119 | # 4 Print out steps in each site No |
120 | # 5 Print character at all nodes of tree No |
121 | # 6 Write out trees onto tree file? Yes |
122 | print CMD "Y\n"; |
123 | close CMD; |
124 | |
125 | # And then we run the program. |
126 | ### HACKY HACKY |
127 | my $PHYLIP_PATH = '/Users/tla/Projects/phylip-3.69/exe'; |
128 | my $program = "pars"; |
129 | if( $^O eq 'darwin' ) { |
130 | $program = "$PHYLIP_PATH/$program.app/Contents/MacOS/$program"; |
131 | } else { |
132 | $program = "$PHYLIP_PATH/$program"; |
133 | } |
134 | |
135 | { |
136 | # We need to run it in our temporary directory where we have created |
137 | # all the expected files. |
138 | local $CWD = $phylip_dir; |
139 | my @cmd = ( $program ); |
140 | run \@cmd, '<', 'cmdfile', '>', '/dev/null'; |
141 | } |
142 | # Now our output should be in 'outfile' and our tree in 'outtree', |
143 | # both in the temp directory. |
144 | |
145 | my @outtree; |
146 | if( -f "$phylip_dir/outtree" ) { |
147 | open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read"; |
148 | @outtree = <TREE>; |
149 | close TREE; |
150 | } |
151 | return( 1, join( '', @outtree ) ) if @outtree; |
152 | |
153 | my @error; |
154 | if( -f "$phylip_dir/output" ) { |
155 | open( OUTPUT, "$phylip_dir/output" ) or die "Could not open output for read"; |
156 | @error = <OUTPUT>; |
157 | close OUTPUT; |
158 | } else { |
159 | push( @error, "Neither outtree nor output file was produced!" ); |
160 | } |
161 | return( undef, join( '', @error ) ); |
162 | } |
163 | |
164 | no Moose; |
165 | __PACKAGE__->meta->make_immutable; |
166 | |
167 | 1; |