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 ) { |
910a0a6d |
25 | warn "Need a linear graph in order to make an alignment table"; |
26 | return; |
9463b0bf |
27 | } |
910a0a6d |
28 | my $table = $self->collation->make_alignment_table; |
29 | # Push the names of the witnesses to initialize the rows of the matrix. |
30 | my @matrix = map { [ $self->_normalize_ac( $_ ) ] } @{$table->[0]}; |
31 | $DB::single = 1; |
32 | foreach my $token_index ( 1 .. $#{$table} ) { |
33 | # First implementation: make dumb alignment table, caring about |
34 | # nothing except which reading is in which position. |
35 | my @chars = convert_characters( $table->[$token_index] ); |
36 | foreach my $idx ( 0 .. $#matrix ) { |
37 | push( @{$matrix[$idx]}, $chars[$idx] ); |
38 | } |
9463b0bf |
39 | } |
910a0a6d |
40 | $self->_save_character_matrix( \@matrix ); |
41 | } |
9463b0bf |
42 | |
910a0a6d |
43 | sub _normalize_ac { |
44 | my( $self, $witname ) = @_; |
45 | my $ac = $self->collation->ac_label; |
46 | if( $witname =~ /(.*)\Q$ac\E$/ ) { |
47 | $witname = $1 . '_ac'; |
9463b0bf |
48 | } |
910a0a6d |
49 | return sprintf( "%-10s", $witname ); |
9463b0bf |
50 | } |
9463b0bf |
51 | |
910a0a6d |
52 | sub convert_characters { |
53 | my $row = shift; |
9463b0bf |
54 | # This is a simple algorithm that treats every reading as different. |
55 | # Eventually we will want to be able to specify how relationships |
56 | # affect the character matrix. |
910a0a6d |
57 | my %unique = ( '__UNDEF__' => 'X' ); |
58 | my $ctr = 0; |
59 | foreach my $word ( @$row ) { |
60 | if( $word && !exists $unique{$word} ) { |
61 | $unique{$word} = chr( 65 + $ctr ); |
62 | $ctr++; |
63 | } |
9463b0bf |
64 | } |
910a0a6d |
65 | if( scalar( keys %unique ) > 8 ) { |
66 | warn "Have more than 8 variants on this location; pars will break"; |
67 | } |
68 | my @chars = map { $_ ? $unique{$_} : $unique{'__UNDEF__' } } @$row; |
69 | return @chars; |
9463b0bf |
70 | } |
71 | |
f6066bac |
72 | sub pars_input { |
9463b0bf |
73 | my $self = shift; |
74 | $self->make_character_matrix unless $self->has_character_matrix; |
f6066bac |
75 | my $matrix = ''; |
76 | my $rows = scalar @{$self->character_matrix}; |
77 | my $columns = scalar @{$self->character_matrix->[0]} - 1; |
78 | $matrix .= "\t$rows\t$columns\n"; |
79 | foreach my $row ( @{$self->character_matrix} ) { |
910a0a6d |
80 | $matrix .= join( '', @$row ) . "\n"; |
f6066bac |
81 | } |
82 | return $matrix; |
83 | } |
84 | |
85 | sub run_pars { |
86 | my $self = shift; |
9463b0bf |
87 | |
88 | # Set up a temporary directory for all the default Phylip files. |
89 | my $phylip_dir = File::Temp->newdir(); |
f6066bac |
90 | # We need an infile, and we need a command input file. |
9463b0bf |
91 | open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile"; |
f6066bac |
92 | print MATRIX $self->pars_input(); |
9463b0bf |
93 | close MATRIX; |
94 | |
95 | open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile"; |
96 | ## TODO any configuration parameters we want to set here |
97 | # U Search for best tree? Yes |
98 | # S Search option? More thorough search |
99 | # V Number of trees to save? 100 |
100 | # J Randomize input order of species? No. Use input order |
101 | # O Outgroup root? No, use as outgroup species 1 |
102 | # T Use Threshold parsimony? No, use ordinary parsimony |
103 | # W Sites weighted? No |
104 | # M Analyze multiple data sets? No |
105 | # I Input species interleaved? Yes |
106 | # 0 Terminal type (IBM PC, ANSI, none)? ANSI |
107 | # 1 Print out the data at start of run No |
108 | # 2 Print indications of progress of run Yes |
109 | # 3 Print out tree Yes |
110 | # 4 Print out steps in each site No |
111 | # 5 Print character at all nodes of tree No |
112 | # 6 Write out trees onto tree file? Yes |
113 | print CMD "Y\n"; |
114 | close CMD; |
115 | |
116 | # And then we run the program. |
117 | ### HACKY HACKY |
118 | my $PHYLIP_PATH = '/Users/tla/Projects/phylip-3.69/exe'; |
119 | my $program = "pars"; |
120 | if( $^O eq 'darwin' ) { |
910a0a6d |
121 | $program = "$PHYLIP_PATH/$program.app/Contents/MacOS/$program"; |
9463b0bf |
122 | } else { |
910a0a6d |
123 | $program = "$PHYLIP_PATH/$program"; |
9463b0bf |
124 | } |
125 | |
126 | { |
910a0a6d |
127 | # We need to run it in our temporary directory where we have created |
128 | # all the expected files. |
129 | local $CWD = $phylip_dir; |
130 | my @cmd = ( $program ); |
131 | run \@cmd, '<', 'cmdfile', '>', '/dev/null'; |
9463b0bf |
132 | } |
133 | # Now our output should be in 'outfile' and our tree in 'outtree', |
134 | # both in the temp directory. |
135 | |
136 | my @outtree; |
137 | if( -f "$phylip_dir/outtree" ) { |
910a0a6d |
138 | open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read"; |
139 | @outtree = <TREE>; |
140 | close TREE; |
9463b0bf |
141 | } |
142 | return( 1, join( '', @outtree ) ) if @outtree; |
143 | |
144 | my @error; |
f6066bac |
145 | if( -f "$phylip_dir/outfile" ) { |
910a0a6d |
146 | open( OUTPUT, "$phylip_dir/outfile" ) or die "Could not open output for read"; |
147 | @error = <OUTPUT>; |
148 | close OUTPUT; |
9463b0bf |
149 | } else { |
910a0a6d |
150 | push( @error, "Neither outtree nor output file was produced!" ); |
9463b0bf |
151 | } |
152 | return( undef, join( '', @error ) ); |
153 | } |
154 | |
155 | no Moose; |
156 | __PACKAGE__->meta->make_immutable; |
157 | |
158 | 1; |