[scpubgit/stemmatology.git] / lib / Text / Tradition / Stemma.pm

package Text::Tradition::Stemma;

use Bio::Phylo::IO;
use Encode qw( decode_utf8 );
use File::chdir;
use File::Temp;
use File::Which;
use Graph;
use Graph::Reader::Dot;
use IPC::Run qw/ run binary /;
use Text::Tradition::StemmaUtil qw/ phylip_pars_input /;
use Moose;

has collation => (
    is => 'ro',
    isa => 'Text::Tradition::Collation',
    required => 1,
    weak_ref => 1,
    );  

has graph => (
    is => 'rw',
    isa => 'Graph',
    predicate => 'has_graph',
    );
    
has distance_trees => (
    is => 'ro',
    isa => 'ArrayRef[Graph]',
    writer => '_save_distance_trees',
    predicate => 'has_distance_trees',
    );
    
has distance_program => (
	is => 'rw',
	isa => 'Str',
	default => '',
	);
    
sub BUILD {
    my( $self, $args ) = @_;
    # If we have been handed a dotfile, initialize it into a graph.
    if( exists $args->{'dot'} ) {
        $self->graph_from_dot( $args->{'dot'} );
    }
}

sub graph_from_dot {
	my( $self, $dotfh ) = @_;
	# Assume utf-8
	binmode( $dotfh, ':utf8' );
 	my $reader = Graph::Reader::Dot->new();
	my $graph = $reader->read_graph( $dotfh );
	if( $graph ) {
		$self->graph( $graph );
		# Go through the nodes and set any non-hypothetical node to extant.
		foreach my $v ( $self->graph->vertices ) {
			$self->graph->set_vertex_attribute( $v, 'class', 'extant' )
				unless $self->graph->has_vertex_attribute( $v, 'class' );
		}
	} else {
		warn "Failed to parse dot in $dotfh";
	}
}

sub as_dot {
    my( $self, $opts ) = @_;
    
    # Get default and specified options
    my %graphopts = ();
    my %nodeopts = (
		'fontsize' => 11,
		'hshape' => 'plaintext',	# Shape for the hypothetical nodes
		'htext' => '*',
		'style' => 'filled',
		'fillcolor' => 'white',
		'shape' => 'ellipse',	# Shape for the extant nodes
	);
	my %edgeopts = (
		'arrowhead' => 'open',
	);
	@graphopts{ keys %{$opts->{'graph'}} } = values %{$opts->{'graph'}} 
		if $opts->{'graph'};
	@nodeopts{ keys %{$opts->{'node'}} } = values %{$opts->{'node'}} 
		if $opts->{'node'};
	@edgeopts{ keys %{$opts->{'edge'}} } = values %{$opts->{'edge'}} 
		if $opts->{'edge'};

	my @dotlines;
	push( @dotlines, 'digraph stemma {' );
	## Print out the global attributes
	push( @dotlines, _make_dotline( 'graph', %graphopts ) ) if keys %graphopts;
	push( @dotlines, _make_dotline( 'edge', %edgeopts ) ) if keys %edgeopts;
	## Delete our special attributes from the node set before continuing
	my $hshape = delete $nodeopts{'hshape'};
	my $htext = delete $nodeopts{'htext'};
	push( @dotlines, _make_dotline( 'node', %nodeopts ) ) if keys %nodeopts;

	# Add each of the nodes.
    foreach my $n ( $self->graph->vertices ) {
        if( $self->graph->get_vertex_attribute( $n, 'class' ) eq 'hypothetical' ) {
        	# Apply our display settings for hypothetical nodes.
        	push( @dotlines, _make_dotline( $n, 'shape' => $hshape, 'label' => $htext ) );
        } else {
        	# Use the default display settings.
            push( @dotlines, "  $n;" );
        }
    }
    # Add each of our edges.
    foreach my $e ( $self->graph->edges ) {
    	my( $from, $to ) = @$e;
    	push( @dotlines, "  $from -> $to;" );
    }
    push( @dotlines, '}' );
    
    return join( "\n", @dotlines );
}


# Another version of dot output meant for graph editing, thus
# much simpler.
sub editable {
	my $self = shift;
	my @dotlines;
	push( @dotlines, 'digraph stemma {' );
	my @real; # A cheap sort
    foreach my $n ( sort $self->graph->vertices ) {
    	my $c = $self->graph->get_vertex_attribute( $n, 'class' );
    	$c = 'extant' unless $c;
    	if( $c eq 'extant' ) {
    		push( @real, $n );
    	} else {
			push( @dotlines, _make_dotline( $n, 'class' => $c ) );
		}
    }
	# Now do the real ones
	foreach my $n ( @real ) {
		push( @dotlines, _make_dotline( $n, 'class' => 'extant' ) );
	}
	foreach my $e ( sort _by_vertex $self->graph->edges ) {
		my( $from, $to ) = @$e;
		push( @dotlines, "  $from -> $to;" );
	}
    push( @dotlines, '}' );
    return join( "\n", @dotlines );
}

sub _make_dotline {
	my( $obj, %attr ) = @_;
	my @pairs;
	foreach my $k ( keys %attr ) {
		my $v = $attr{$k};
		$v =~ s/\"/\\\"/g;
		push( @pairs, "$k=\"$v\"" );
	}
	return sprintf( "  %s [ %s ];", $obj, join( ', ', @pairs ) );
}
	
sub _by_vertex {
	return $a->[0].$a->[1] cmp $b->[0].$b->[1];
}

# Render the stemma as SVG.
sub as_svg {
    my( $self, $opts ) = @_;
    my $dot = $self->as_dot( $opts );
    my @cmd = qw/dot -Tsvg/;
    my( $svg, $err );
    my $dotfile = File::Temp->new();
    ## TODO REMOVE
    # $dotfile->unlink_on_destroy(0);
    binmode $dotfile, ':utf8';
    print $dotfile $dot;
    push( @cmd, $dotfile->filename );
    run( \@cmd, ">", binary(), \$svg );
    $svg = decode_utf8( $svg );
    return $svg;
}

sub witnesses {
    my $self = shift;
    my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' }
        $self->graph->vertices;
    return @wits;
}

#### Methods for calculating phylogenetic trees ####

before 'distance_trees' => sub {
    my $self = shift;
    my %args = (
    	'program' => 'phylip_pars',
    	@_ );
    # TODO allow specification of method for calculating distance tree
    if( !$self->has_distance_trees
    	|| $args{'program'} ne $self->distance_program ) {
        # We need to make a tree before we can return it.
        my $dsub = 'run_' . $args{'program'};
        my( $ok, $result ) = $self->$dsub();
        if( $ok ) {
            # Save the resulting trees
            my $trees = _parse_newick( $result );
            $self->_save_distance_trees( $trees );
            $self->distance_program( $args{'program'} );
        } else {
            warn "Failed to calculate distance trees: $result";
        }
    }
};

sub run_phylip_pars {
    my $self = shift;

    # Set up a temporary directory for all the default Phylip files.
    my $phylip_dir = File::Temp->newdir();
    # $phylip_dir->unlink_on_destroy(0);
    # We need an infile, and we need a command input file.
    open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile";
    print MATRIX phylip_pars_input( $self->collation->make_alignment_table() );
    close MATRIX;

    open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile";
    ## TODO any configuration parameters we want to set here
#   U                 Search for best tree?  Yes
#   S                        Search option?  More thorough search
#   V              Number of trees to save?  100
#   J     Randomize input order of species?  No. Use input order
#   O                        Outgroup root?  No, use as outgroup species 1
#   T              Use Threshold parsimony?  No, use ordinary parsimony
#   W                       Sites weighted?  No
#   M           Analyze multiple data sets?  No
#   I            Input species interleaved?  Yes
#   0   Terminal type (IBM PC, ANSI, none)?  ANSI
#   1    Print out the data at start of run  No
#   2  Print indications of progress of run  Yes
#   3                        Print out tree  Yes
#   4          Print out steps in each site  No
#   5  Print character at all nodes of tree  No
#   6       Write out trees onto tree file?  Yes
    print CMD "Y\n";
    close CMD;

    # And then we run the program.
    my $program = File::Which::which( 'pars' );
    unless( -x $program ) {
		return( undef, "Phylip pars not found in path" );
    }

    {
        # We need to run it in our temporary directory where we have created
        # all the expected files.
        local $CWD = $phylip_dir;
        my @cmd = ( $program );
        run \@cmd, '<', 'cmdfile', '>', '/dev/null';
    }
    # Now our output should be in 'outfile' and our tree in 'outtree',
    # both in the temp directory.

    my @outtree;
    if( -f "$phylip_dir/outtree" ) {
        open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read";
        @outtree = <TREE>;
        close TREE;
    }
    return( 1, join( '', @outtree ) ) if @outtree;

    my @error;
    if( -f "$phylip_dir/outfile" ) {
        open( OUTPUT, "$phylip_dir/outfile" ) or die "Could not open output for read";
        @error = <OUTPUT>;
        close OUTPUT;
    } else {
        push( @error, "Neither outtree nor output file was produced!" );
    }
    return( undef, join( '', @error ) );
}

sub _parse_newick {
    my $newick = shift;
    my @trees;
    # Parse the result into a tree
    my $forest = Bio::Phylo::IO->parse( 
        -format => 'newick',
        -string => $newick,
        );
    # Turn the tree into a graph, starting with the root node
    foreach my $tree ( @{$forest->get_entities} ) {
        push( @trees, _graph_from_bio( $tree ) );
    }
    return \@trees;
}

sub _graph_from_bio {
    my $tree = shift;
    my $graph = Graph->new( 'undirected' => 1 );
    # Give all the intermediate anonymous nodes a name.
    my $i = 0;
    foreach my $n ( @{$tree->get_entities} ) {
        next if $n->get_name;
        $n->set_name( $i++ );
    }
    my $root = $tree->get_root->get_name;
    $graph->add_vertex( $root );
    _add_tree_children( $graph, $root, $tree->get_root->get_children() );
    return $graph;
}

sub _add_tree_children {
    my( $graph, $parent, $tree_children ) = @_;
    foreach my $c ( @$tree_children ) {
        my $child = $c->get_name;
        $graph->add_vertex( $child );
        $graph->add_path( $parent, $child );
        _add_tree_children( $graph, $child, $c->get_children() );
    }
}

no Moose;
__PACKAGE__->meta->make_immutable;
    
1;
Commit	Line	Data
9463b0bf	1	package Text::Tradition::Stemma;
9463b0bf	2
40f19742	3	use Bio::Phylo::IO;
e79c23c7	4	use Encode qw( decode_utf8 );
9463b0bf	5	use File::chdir;
9463b0bf	6	use File::Temp;
457b1620	7	use File::Which;
e05997e2	8	use Graph;
e05997e2	9	use Graph::Reader::Dot;
e79c23c7	10	use IPC::Run qw/ run binary /;
68454b71	11	use Text::Tradition::StemmaUtil qw/ phylip_pars_input /;
40f19742	12	use Moose;
9463b0bf	13
	14	has collation => (
	15	is => 'ro',
	16	isa => 'Text::Tradition::Collation',
	17	required => 1,
8d9a1cd8	18	weak_ref => 1,
9463b0bf	19	);
9463b0bf	20
e05997e2	21	has graph => (
	22	is => 'rw',
	23	isa => 'Graph',
	24	predicate => 'has_graph',
	25	);
	26
40f19742	27	has distance_trees => (
	28	is => 'ro',
	29	isa => 'ArrayRef[Graph]',
	30	writer => '_save_distance_trees',
	31	predicate => 'has_distance_trees',
	32	);
c0ccdb62	33
0f5d05c6	34	has distance_program => (
	35	is => 'rw',
	36	isa => 'Str',
	37	default => '',
	38	);
	39
e05997e2	40	sub BUILD {
	41	my( $self, $args ) = @_;
	42	# If we have been handed a dotfile, initialize it into a graph.
	43	if( exists $args->{'dot'} ) {
8d9a1cd8	44	$self->graph_from_dot( $args->{'dot'} );
e05997e2	45	}
c0ccdb62	46	}
c0ccdb62	47
8d9a1cd8	48	sub graph_from_dot {
	49	my( $self, $dotfh ) = @_;
	50	# Assume utf-8
	51	binmode( $dotfh, ':utf8' );
	52	my $reader = Graph::Reader::Dot->new();
	53	my $graph = $reader->read_graph( $dotfh );
7a7c249c	54	if( $graph ) {
	55	$self->graph( $graph );
	56	# Go through the nodes and set any non-hypothetical node to extant.
	57	foreach my $v ( $self->graph->vertices ) {
	58	$self->graph->set_vertex_attribute( $v, 'class', 'extant' )
	59	unless $self->graph->has_vertex_attribute( $v, 'class' );
	60	}
	61	} else {
	62	warn "Failed to parse dot in $dotfh";
	63	}
8d9a1cd8	64	}
	65
	66	sub as_dot {
e367f5c0	67	my( $self, $opts ) = @_;
7a7c249c	68
	69	# Get default and specified options
	70	my %graphopts = ();
	71	my %nodeopts = (
	72	'fontsize' => 11,
	73	'hshape' => 'plaintext', # Shape for the hypothetical nodes
	74	'htext' => '*',
	75	'style' => 'filled',
	76	'fillcolor' => 'white',
	77	'shape' => 'ellipse', # Shape for the extant nodes
	78	);
	79	my %edgeopts = (
	80	'arrowhead' => 'open',
	81	);
	82	@graphopts{ keys %{$opts->{'graph'}} } = values %{$opts->{'graph'}}
	83	if $opts->{'graph'};
	84	@nodeopts{ keys %{$opts->{'node'}} } = values %{$opts->{'node'}}
	85	if $opts->{'node'};
	86	@edgeopts{ keys %{$opts->{'edge'}} } = values %{$opts->{'edge'}}
	87	if $opts->{'edge'};
	88
	89	my @dotlines;
	90	push( @dotlines, 'digraph stemma {' );
	91	## Print out the global attributes
	92	push( @dotlines, _make_dotline( 'graph', %graphopts ) ) if keys %graphopts;
	93	push( @dotlines, _make_dotline( 'edge', %edgeopts ) ) if keys %edgeopts;
	94	## Delete our special attributes from the node set before continuing
	95	my $hshape = delete $nodeopts{'hshape'};
	96	my $htext = delete $nodeopts{'htext'};
	97	push( @dotlines, _make_dotline( 'node', %nodeopts ) ) if keys %nodeopts;
	98
	99	# Add each of the nodes.
	100	foreach my $n ( $self->graph->vertices ) {
	101	if( $self->graph->get_vertex_attribute( $n, 'class' ) eq 'hypothetical' ) {
	102	# Apply our display settings for hypothetical nodes.
	103	push( @dotlines, _make_dotline( $n, 'shape' => $hshape, 'label' => $htext ) );
e79c23c7	104	} else {
7a7c249c	105	# Use the default display settings.
7a7c249c	106	push( @dotlines, " $n;" );
e79c23c7	107	}
e79c23c7	108	}
7a7c249c	109	# Add each of our edges.
	110	foreach my $e ( $self->graph->edges ) {
	111	my( $from, $to ) = @$e;
	112	push( @dotlines, " $from -> $to;" );
	113	}
	114	push( @dotlines, '}' );
e79c23c7	115
7a7c249c	116	return join( "\n", @dotlines );
	117	}
	118
	119
	120	# Another version of dot output meant for graph editing, thus
	121	# much simpler.
	122	sub editable {
	123	my $self = shift;
	124	my @dotlines;
	125	push( @dotlines, 'digraph stemma {' );
	126	my @real; # A cheap sort
	127	foreach my $n ( sort $self->graph->vertices ) {
	128	my $c = $self->graph->get_vertex_attribute( $n, 'class' );
	129	$c = 'extant' unless $c;
	130	if( $c eq 'extant' ) {
	131	push( @real, $n );
	132	} else {
	133	push( @dotlines, _make_dotline( $n, 'class' => $c ) );
	134	}
e367f5c0	135	}
7a7c249c	136	# Now do the real ones
	137	foreach my $n ( @real ) {
	138	push( @dotlines, _make_dotline( $n, 'class' => 'extant' ) );
	139	}
	140	foreach my $e ( sort _by_vertex $self->graph->edges ) {
	141	my( $from, $to ) = @$e;
	142	push( @dotlines, " $from -> $to;" );
	143	}
	144	push( @dotlines, '}' );
	145	return join( "\n", @dotlines );
	146	}
	147
	148	sub _make_dotline {
	149	my( $obj, %attr ) = @_;
	150	my @pairs;
	151	foreach my $k ( keys %attr ) {
	152	my $v = $attr{$k};
	153	$v =~ s/\"/\\\"/g;
	154	push( @pairs, "$k=\"$v\"" );
	155	}
	156	return sprintf( " %s [ %s ];", $obj, join( ', ', @pairs ) );
8d9a1cd8	157	}
8d9a1cd8	158
7a7c249c	159	sub _by_vertex {
	160	return $a->[0].$a->[1] cmp $b->[0].$b->[1];
	161	}
8d9a1cd8	162
	163	# Render the stemma as SVG.
	164	sub as_svg {
	165	my( $self, $opts ) = @_;
	166	my $dot = $self->as_dot( $opts );
e79c23c7	167	my @cmd = qw/dot -Tsvg/;
	168	my( $svg, $err );
	169	my $dotfile = File::Temp->new();
	170	## TODO REMOVE
	171	# $dotfile->unlink_on_destroy(0);
	172	binmode $dotfile, ':utf8';
8d9a1cd8	173	print $dotfile $dot;
e79c23c7	174	push( @cmd, $dotfile->filename );
	175	run( \@cmd, ">", binary(), \$svg );
	176	$svg = decode_utf8( $svg );
	177	return $svg;
	178	}
	179
08e0fb85	180	sub witnesses {
	181	my $self = shift;
	182	my @wits = grep { $self->graph->get_vertex_attribute( $_, 'class' ) eq 'extant' }
	183	$self->graph->vertices;
	184	return @wits;
	185	}
	186
e79c23c7	187	#### Methods for calculating phylogenetic trees ####
e79c23c7	188
40f19742	189	before 'distance_trees' => sub {
40f19742	190	my $self = shift;
0f5d05c6	191	my %args = (
	192	'program' => 'phylip_pars',
	193	@_ );
40f19742	194	# TODO allow specification of method for calculating distance tree
0f5d05c6	195	if( !$self->has_distance_trees
0f5d05c6	196	\|\| $args{'program'} ne $self->distance_program ) {
40f19742	197	# We need to make a tree before we can return it.
0f5d05c6	198	my $dsub = 'run_' . $args{'program'};
0f5d05c6	199	my( $ok, $result ) = $self->$dsub();
40f19742	200	if( $ok ) {
c0ccdb62	201	# Save the resulting trees
	202	my $trees = _parse_newick( $result );
	203	$self->_save_distance_trees( $trees );
0f5d05c6	204	$self->distance_program( $args{'program'} );
40f19742	205	} else {
c0ccdb62	206	warn "Failed to calculate distance trees: $result";
40f19742	207	}
	208	}
	209	};
f6066bac	210
40f19742	211	sub run_phylip_pars {
f6066bac	212	my $self = shift;
9463b0bf	213
	214	# Set up a temporary directory for all the default Phylip files.
	215	my $phylip_dir = File::Temp->newdir();
eca16057	216	# $phylip_dir->unlink_on_destroy(0);
f6066bac	217	# We need an infile, and we need a command input file.
9463b0bf	218	open( MATRIX, ">$phylip_dir/infile" ) or die "Could not write $phylip_dir/infile";
68454b71	219	print MATRIX phylip_pars_input( $self->collation->make_alignment_table() );
9463b0bf	220	close MATRIX;
	221
	222	open( CMD, ">$phylip_dir/cmdfile" ) or die "Could not write $phylip_dir/cmdfile";
	223	## TODO any configuration parameters we want to set here
	224	# U Search for best tree? Yes
	225	# S Search option? More thorough search
	226	# V Number of trees to save? 100
	227	# J Randomize input order of species? No. Use input order
	228	# O Outgroup root? No, use as outgroup species 1
	229	# T Use Threshold parsimony? No, use ordinary parsimony
	230	# W Sites weighted? No
	231	# M Analyze multiple data sets? No
	232	# I Input species interleaved? Yes
	233	# 0 Terminal type (IBM PC, ANSI, none)? ANSI
	234	# 1 Print out the data at start of run No
	235	# 2 Print indications of progress of run Yes
	236	# 3 Print out tree Yes
	237	# 4 Print out steps in each site No
	238	# 5 Print character at all nodes of tree No
	239	# 6 Write out trees onto tree file? Yes
	240	print CMD "Y\n";
	241	close CMD;
	242
	243	# And then we run the program.
457b1620	244	my $program = File::Which::which( 'pars' );
457b1620	245	unless( -x $program ) {
0f5d05c6	246	return( undef, "Phylip pars not found in path" );
9463b0bf	247	}
	248
	249	{
910a0a6d	250	# We need to run it in our temporary directory where we have created
	251	# all the expected files.
	252	local $CWD = $phylip_dir;
	253	my @cmd = ( $program );
	254	run \@cmd, '<', 'cmdfile', '>', '/dev/null';
9463b0bf	255	}
	256	# Now our output should be in 'outfile' and our tree in 'outtree',
	257	# both in the temp directory.
	258
	259	my @outtree;
	260	if( -f "$phylip_dir/outtree" ) {
910a0a6d	261	open( TREE, "$phylip_dir/outtree" ) or die "Could not open outtree for read";
	262	@outtree = <TREE>;
	263	close TREE;
9463b0bf	264	}
	265	return( 1, join( '', @outtree ) ) if @outtree;
	266
	267	my @error;
f6066bac	268	if( -f "$phylip_dir/outfile" ) {
910a0a6d	269	open( OUTPUT, "$phylip_dir/outfile" ) or die "Could not open output for read";
	270	@error = <OUTPUT>;
	271	close OUTPUT;
9463b0bf	272	} else {
910a0a6d	273	push( @error, "Neither outtree nor output file was produced!" );
9463b0bf	274	}
	275	return( undef, join( '', @error ) );
	276	}
	277
40f19742	278	sub _parse_newick {
	279	my $newick = shift;
	280	my @trees;
	281	# Parse the result into a tree
	282	my $forest = Bio::Phylo::IO->parse(
	283	-format => 'newick',
	284	-string => $newick,
	285	);
	286	# Turn the tree into a graph, starting with the root node
	287	foreach my $tree ( @{$forest->get_entities} ) {
	288	push( @trees, _graph_from_bio( $tree ) );
	289	}
	290	return \@trees;
	291	}
	292
	293	sub _graph_from_bio {
	294	my $tree = shift;
	295	my $graph = Graph->new( 'undirected' => 1 );
	296	# Give all the intermediate anonymous nodes a name.
	297	my $i = 0;
	298	foreach my $n ( @{$tree->get_entities} ) {
	299	next if $n->get_name;
	300	$n->set_name( $i++ );
	301	}
	302	my $root = $tree->get_root->get_name;
	303	$graph->add_vertex( $root );
	304	_add_tree_children( $graph, $root, $tree->get_root->get_children() );
	305	return $graph;
	306	}
	307
	308	sub _add_tree_children {
	309	my( $graph, $parent, $tree_children ) = @_;
	310	foreach my $c ( @$tree_children ) {
	311	my $child = $c->get_name;
	312	$graph->add_vertex( $child );
	313	$graph->add_path( $parent, $child );
	314	_add_tree_children( $graph, $child, $c->get_children() );
	315	}
	316	}
	317
9463b0bf	318	no Moose;
	319	__PACKAGE__->meta->make_immutable;
	320
	321	1;