| Bio::Graphics::Panel - Generate GD images of Bio::Seq objects |
Bio::Graphics::Panel - Generate GD images of Bio::Seq objects
# This script parses a GenBank or EMBL file named on the command # line and produces a PNG rendering of it. Call it like this: # render.pl my_file.embl | display -
use strict; use Bio::Graphics; use Bio::SeqIO;
my $file = shift or die "provide a sequence file as the argument"; my $io = Bio::SeqIO->new(-file=>$file) or die "could not create Bio::SeqIO"; my $seq = $io->next_seq or die "could not find a sequence in the file";
my @features = $seq->all_SeqFeatures;
# sort features by their primary tags my %sorted_features; for my $f (@features) { my $tag = $f->primary_tag; push @{$sorted_features{$tag}},$f; }
my $panel = Bio::Graphics::Panel->new( -length => $seq->length, -key_style => 'between', -width => 800, -pad_left => 10, -pad_right => 10, ); $panel->add_track( arrow => Bio::SeqFeature::Generic->new(-start=>1, -end=>$seq->length), -bump => 0, -double=>1, -tick => 2); $panel->add_track(generic => Bio::SeqFeature::Generic->new(-start=>1, -end=>$seq->length), -glyph => 'generic', -bgcolor => 'blue', -label => 1, );
# general case my @colors = qw(cyan orange blue purple green chartreuse magenta yellow aqua); my $idx = 0; for my $tag (sort keys %sorted_features) { my $features = $sorted_features{$tag}; $panel->add_track($features, -glyph => 'generic', -bgcolor => $colors[$idx++ % @colors], -fgcolor => 'black', -font2color => 'red', -key => "${tag}s", -bump => +1, -height => 8, -label => 1, -description => 1, ); }
print $panel->png; $panel->finished;
exit 0;
The Bio::Graphics::Panel class provides drawing and formatting services for any object that implements the Bio::SeqFeatureI interface, including Ace::Sequence::Feature and Das::Segment::Feature objects. It can be used to draw sequence annotations, physical (contig) maps, or any other type of map in which a set of discrete ranges need to be laid out on the number line.
The module supports a drawing style in which each type of feature occupies a discrete "track" that spans the width of the display. Each track will have its own distinctive "glyph", a configurable graphical representation of the feature.
The module also supports a more flexible style in which several different feature types and their associated glyphs can occupy the same track. The choice of glyph is under run-time control.
Semantic zooming (for instance, changing the type of glyph depending on the density of features) is supported by a callback system for configuration variables. The module has built-in support for Bio::Das stylesheets, and stylesheet-driven configuration can be intermixed with semantic zooming, if desired.
You can add a key to the generated image using either of two key styles. One style places the key captions at the top of each track. The other style generates a graphical key at the bottom of the image.
Note that this module depends on GD. The optional SVG output depends on GD::SVG and SVG.
The installed script glyph_help.pl provides quick help on glyphs and their options.
This section describes the class and object methods for Bio::Graphics::Panel.
Typically you will begin by creating a new Bio::Graphics::Panel
object, passing it the desired width of the image to generate and an
origin and length describing the coordinate range to display. The
Bio::Graphics::Panel->new() method has many configuration variables
that allow you to control the appearance of the image.
You will then call add_track() one or more times to add sets of
related features to the picture. add_track() places a new horizontal
track on the image, and is likewise highly configurable. When you
have added all the features you desire, you may call png() to convert
the image into a PNG-format image, or boxes() to return coordinate
information that can be used to create an imagemap.
new() is the constructor for Bio::Graphics::Panel:
new(@options)
The new() method creates a new panel object. The options are
a set of tag/value pairs as follows:
Option Value Default ------ ----- -------
-offset Base pair to place at extreme left none
of image, in zero-based coordinates
-length Length of sequence segment, in bp none
-start Start of range, in 1-based none
coordinates.
-stop Stop of range, in 1-based none
coordinates.
-end Same as -stop.
-segment A Bio::SeqI or Das::Segment none
object, used to derive sequence
range if not otherwise specified.
-width Desired width of image, in pixels 600
-spacing Spacing between tracks, in pixels 5
-pad_top Additional whitespace between top 0
of image and contents, in pixels
-pad_bottom Additional whitespace between top 0
of image and bottom, in pixels
-pad_left Additional whitespace between left 0
of image and contents, in pixels
-pad_right Additional whitespace between right 0
of image and bottom, in pixels
-bgcolor Background color for the panel as a white
whole
-key_color Background color for the key printed wheat
at bottom of panel (if any)
-key_spacing Spacing between key glyphs in the 10
key printed at bottom of panel
(if any)
-key_font Font to use in printed key gdMediumBoldFont
captions.
-key_style Whether to print key at bottom of none
panel ("bottom"), between each
track ("between"), to the left of
each track ("left"), to the right
of each track ("right") or
not at all ("none").
-add_category_labels false
Whether to add the "category" to
the track key. The category is
an optional argument that can
be attached to each track. If
a category is present, and this
option is true, then the category
will be added to the track label
in parentheses. For example, if
-key is "Protein matches" and
-category is "vertebrate", then
the track will be labeled
"Protein matches (vertebrate)".
-auto_pad If "left" or "right" keys are in use true
then setting auto_pad to a true value
will allow the panel to adjust its
width in order to accomodate the
length of the longest key.
-empty_tracks What to do when a track is empty. suppress
Options are to suppress the track
completely ("suppress"), to show just
the key in "between" mode ("key"),
to draw a thin grey line ("line"),
or to draw a dashed line ("dashed").
-flip flip the drawing coordinates left false
to right, so that lower coordinates
are to the right. This can be
useful for drawing (-) strand
features.
-all_callbacks Whether to invoke callbacks on false
the automatic "track" and "group"
glyphs.
-grid Whether to draw a vertical grid in false
the background. Pass a scalar true
value to have a grid drawn at
regular intervals (corresponding
to the minor ticks of the arrow
glyph). Pass an array reference
to draw the grid at the specified
positions.
-gridcolor Color of the grid lightcyan
-gridmajorcolor Color of grid major intervals cyan
-extend_grid If true, extend the grid into the pad false
top and pad_bottom regions
-background An image or callback to use for the none
background of the image. Will be
invoked I<before> drawing the grid.
-postgrid An image or callback to use for the none
background of the image. Will be
invoked I<after> drawing the grid.
-truecolor Create a truecolor (24-bit) image. false
Useful when working with the
"image" glyph.
-image_class To create output in scalable vector
graphics (SVG), optionally pass the image
class parameter 'GD::SVG'. Defaults to
using vanilla GD. See the corresponding
image_class() method below for details.
-link, -title, -target
These options are used when creating imagemaps
for display on the web. See L</"Creating Imagemaps">.
Typically you will pass new() an object that implements the
Bio::RangeI interface, providing a length() method, from which the
panel will derive its scale.
$panel = Bio::Graphics::Panel->new(-segment => $sequence, -width => 800);
new() will return undef in case of an error.
Note that if you use the "left" or "right" key styles, you are responsible for allocating sufficient -pad_left or -pad_right room for the labels to appear. The necessary width is the number of characters in the longest key times the font width (gdMediumBoldFont by default) plus 3 pixels of internal padding. The simplest way to calculate this is to iterate over the possible track labels, find the largest one, and then to compute its width using the formula:
$width = gdMediumBoldFont->width * length($longest_key) +3;
In order to obtain scalable vector graphics (SVG) output, you should
pass new() the -image_class=>'GD::SVG' parameter. This will cause
Bio::Graphics::Panel to load the optional GD::SVG module. See the gd()
and svg() methods below for additional information.
You can tile an image onto the panel either before or after it draws the grid. Simply provide the filename of the image in the -background or -postgrid options. The image file must be of type PNG, JPEG, XBM or GIF and have a filename ending in .png, .jpg, .jpeg, .xbm or .gif.
You can also pass a code ref for the -background or -postgrid option, in which case the subroutine will be invoked at the appropriate time with the GD::Image object and the Panel object as its two arguments. You can then use the panel methods to map base pair coordinates into pixel coordinates and do some custom drawing. For example, this code fragment will draw a gray rectangle between bases 500 and 600 to indicate a "gap" in the sequence:
my $panel = Bio::Graphics::Panel->new(-segment=>$segment, -grid=>1, -width=>600, -postgrid=> \&draw_gap); sub gap_it { my $gd = shift; my $panel = shift; my ($gap_start,$gap_end) = $panel->location2pixel(500,600); my $top = $panel->top; my $bottom = $panel->bottom; my $gray = $panel->translate_color('gray'); $gd->filledRectangle($gap_start,$top,$gap_end,$bottom,$gray); }
add_track($glyph,$features,@options)
The add_track() method adds a new track to the image.
Tracks are horizontal bands which span the entire width of the panel. Each track contains a number of graphical elements called "glyphs", corresponding to a sequence feature.
There are a large number of glyph types. By default, each track will
be homogeneous on a single glyph type, but you can mix several glyph
types on the same track by providing a code reference to the -glyph
argument. Other options passed to add_track() control the color and
size of the glyphs, whether they are allowed to overlap, and other
formatting attributes. The height of a track is determined from its
contents and cannot be directly influenced.
The first two arguments are the glyph name and an array reference containing the list of features to display. The order of the arguments is irrelevant, allowing either of these idioms:
$panel->add_track(arrow => \@features); $panel->add_track(\@features => 'arrow');
The glyph name indicates how each feature is to be rendered. A variety of glyphs are available, and the number is growing. You may omit the glyph name entirely by providing a -glyph argument among @options, as described below.
Currently, the following glyphs are available:
Name Description ---- -----------
anchored_arrow
a span with vertical bases |---------|. If one or
the other end of the feature is off-screen, the base
will be replaced by an arrow.
arrow An arrow; can be unidirectional or bidirectional. It is also capable of displaying a scale with major and minor tickmarks, and can be oriented horizontally or vertically.
box A filled rectangle, nondirectional. Subfeatures are ignored.
cds Draws CDS features, using the phase information to
show the reading frame usage. At high magnifications
draws the protein translation.
crossbox A box with a big "X" inside it.
diamond A diamond, useful for point features like SNPs.
dna At high magnification draws the DNA sequence. At
low magnifications draws the GC content.
dot A circle, useful for point features like SNPs, stop
codons, or promoter elements.
ellipse An oval.
extending_arrow
Similar to arrow, but a dotted line indicates when the
feature extends beyond the end of the canvas.
generic A filled rectangle, nondirectional. Subfeatures are shown
as rectangles that are not connected together.
graded_segments
Similar to segments, but the intensity of the color
is proportional to the score of the feature. This
is used for showing the intensity of blast hits or
other alignment features.
group A group of related features connected by a dashed line.
This is used internally by Panel.
image A pixmap image that will be layered on top of the graphic.
heterogeneous_segments
Like segments, but you can use the source field of the feature
to change the color of each segment.
line A simple line.
pinsertion A triangle designed to look like an insertion location
(e.g. a transposon insertion).
processed_transcript multi-purpose representation of a spliced mRNA, including
positions of UTRs
primers Two inward pointing arrows connected by a line.
Used for STSs.
redgreen_box A box that changes from green->yellow->red as the score
of the feature increases from 0.0 to 1.0. Useful for
representing microarray results.
rndrect A round-cornered rectangle.
segments A set of filled rectangles connected by solid lines.
Used for interrupted features, such as gapped
alignments.
ruler_arrow An arrow with major and minor tick marks and interval
labels.
toomany Tries to show many features as a cloud. Not very successful.
track A group of related features not connected by a line.
This is used internally by Panel.
transcript Similar to segments, but the connecting line is
a "hat" shape, and the direction of transcription
is indicated by a small arrow.
transcript2 Similar to transcript, but the direction of
transcription is indicated by a terminal exon
in the shape of an arrow.
translation 1, 2 and 3-frame translations. At low magnifications,
can be configured to show start and stop codon locations.
At high magnifications, shows the multi-frame protein
translation.
triangle A triangle whose width and orientation can be altered.
xyplot Histograms and other graphs plotted against the genome.
stackedplot A column plot showing multiple data series across multiple categories.
ternary_plot Ternary (triangle) plots.
whiskerplot Box and whisker plot for statistical data
If the glyph name is omitted from add_track(), the "generic" glyph will be used by default. To get more information about a glyph, run perldoc on "Bio::Graphics::Glyph::glyphname", replacing "glyphname" with the name of the glyph you are interested in.
The "box" glyph is optimized for single features with no subfeatures. If you are drawing such a feature, using "box" will be noticeably faster than "generic."
The @options array is a list of name/value pairs that control the attributes of the track. Some options are interpretered directly by the track. Others are passed down to the individual glyphs (see GLYPH OPTIONS). The following options are track-specific:
Option Description Default ------ ----------- -------
-tkcolor Track color white
-glyph Glyph class to use. "generic"
-stylesheet Bio::Das::Stylesheet to none
use to generate glyph
classes and options.
-tkcolor controls the background color of the track as a whole.
-glyph controls the glyph type. If present, it supersedes the
glyph name given in the first or second argument to add_track(). The
value of -glyph may be a constant string, a hash reference, or a
code reference. In the case of a constant string, that string will be
used as the class name for all generated glyphs. If a hash reference
is passed, then the feature's primary_tag() will be used as the key to
the hash, and the value, if any, used to generate the glyph type. If
a code reference is passed, then this callback will be passed
arguments consisting of the feature and the panel object. The
callback is expected to examine the feature and return a glyph name as
its single result.
Example:
$panel->add_track(\@exons, -glyph => sub { my ($feature,$panel) = @_; $feature->source_tag eq 'curated' ? 'ellipse' : 'box'; } );
The -stylesheet argument is used to pass a Bio::Das stylesheet object to the panel. This stylesheet will be called to determine both the glyph and the glyph options. If both a stylesheet and direct options are provided, the latter take precedence.
If successful, add_track() returns an Bio::Graphics::Glyph object.
You can use this object to add additional features or to control the
appearance of the track with greater detail, or just ignore it.
Tracks are added in order from the top of the image to the bottom. To
add tracks to the top of the image, use unshift_track().
Adding groups of features: It is not uncommon to add a group of features which are logically connected, such as the 5' and 3' ends of EST reads. To group features into sets that remain on the same horizontal position and bump together, pass the sets as an anonymous array. For example:
$panel->add_track(segments => [[$abc_5,$abc_3], [$xxx_5,$xxx_3], [$yyy_5,$yyy_3]] );
Typical usage is:
$panel->add_track( transcript => \@genes, -fillcolor => 'green', -fgcolor => 'black', -bump => +1, -height => 10, -label => 1);
The track object is simply a specialized type of glyph. See the Bio::Graphics::Glyph manpage for a description of the methods that it supports.
unshift_track($glyph,$features,@options)
unshift_track() works like add_track(), except that the new track is
added to the top of the image rather than the bottom.
insert_track($position,$glyph,$features,@options)
This works like add_track(), but the track is inserted into the indicated position. The track will be inserted before the indicated position; thus specify a track of 0 to insert the new track at the beginning.
gd([$gd])
The gd() method lays out the image and returns a GD::Image object
containing it. You may then call the GD::Image object's png() or
jpeg() methods to get the image data.
Optionally, you may pass gd() a preexisting GD::Image object that you
wish to draw on top of. If you do so, you should call the width() and
height() methods first to ensure that the image has sufficient
dimensions.
If you passed new() the -image_class=>'GD::SVG' parameter, the gd() method
returns a GD::SVG::Image object. This object overrides GD::Image
methods in order to generate SVG output. It behaves exactly as
described for GD::Image objects with one exception: it implements and
svg() method instead of the png() or jpeg() methods. Currently there
is no direct access to underlying SVG calls but this is subject to
change in the future.
The png() method returns the image as a PNG-format drawing, without
the intermediate step of returning a GD::Image object.
The svg() method returns the image in an XML-ified SVG format.
Bio::Graphics creates memory cycles. When you are finished with the
panel, you should call its finished() method. Otherwise you will have
memory leaks. This is only an issue if you're going to create several
panels in a single program.
The image_class() method returns the current drawing package being
used, currently one of GD or GD::SVG. This is primarily used
internally to ensure that calls to GD's exported methods are called in
an object-oriented manner to avoid compile time undefined string
errors. This is usually not needed for external use.
This accessor method, like image_class() above is provided as a
convenience. It returns the current image package in use, currently
one of GD::Image or GD::SVG::Image. This is not normally used
externally.
This accessor method, like image_package() above is provided as a
convenience. It returns the current polygon package in use, currently
one of GD::Polygon or GD::SVG::Polygon. This is not normally used
externally except in the design of glyphs.
The boxes() method returns a list of arrayrefs containing the
coordinates of each glyph. The method is useful for constructing an
image map. In a scalar context, boxes() returns an arrayref. In an
list context, the method returns the list directly.
Each member of the list is an arrayref of the following format:
[ $feature, $x1, $y1, $x2, $y2, $track ]
The first element is the feature object; either an Ace::Sequence::Feature, a Das::Segment::Feature, or another Bioperl Bio::SeqFeatureI object. The coordinates are the topleft and bottomright corners of the glyph, including any space allocated for labels. The track is the Bio::Graphics::Glyph object corresponding to the track that the feature is rendered inside.
Returns the positions of the track keys as an arrayref or a list, depending on context. Each value in the list is an arrayref of format:
[ $key_text, $x1, $y1, $x2, $y2, $track ]
track_position($track)
After calling gd() or boxes(), you can learn the resulting Y
coordinate of a track by calling track_position() with the value
returned by add_track() or unshift_track(). This will return undef if
called before gd() or boxes() or with an invalid track.
rotate([$new_value])
Gets or sets the "rotate" flag. If rotate is set to true (default
false), then calls to gd(), png(), gif(), boxes(), and image_and_map()
will all return an image and/or imagemap that has been rotated to the
right by 90 degrees. This is mostly useful for drawing karyotypes with
the ideogram glyph, in order to rotate the chromosomes into the usual
vertical position.
location2pixel(@feature_coords)
Public routine to map feature coordinates (in base pairs) into pixel coordinates relative to the left-hand edge of the picture. If you define a -background callback, the callback may wish to invoke this routine in order to translate base coordinates into pixel coordinates.
Return the pixel coordinates of the drawing area of the panel, that is, exclusive of the padding.
Each glyph has its own specialized subset of options, but some are shared by all glyphs:
Option Description Default ------ ----------- -------
-key Description of track for undef
display in the track label.
-category The category of the track undef
for display in the
track label.
-fgcolor Foreground color black
-bgcolor Background color turquoise
-linewidth Width of lines drawn by 1
glyph
-height Height of glyph 10
-font Glyph font gdSmallFont
-fontcolor Primary font color black
-font2color Secondary font color turquoise
-label Whether to draw a label false
-description Whether to draw a false
description
-bump Bump direction 0
-sort_order Specify layout sort order "default"
-feature_limit
Maximum number of features undef (unlimited)
to display
-bump_limit Maximum number of levels undef (unlimited)
to bump
-hbumppad Additional horizontal 0
padding between bumped
features
-strand_arrow Whether to indicate undef (false)
strandedness
-stranded Synonym for -strand_arrow undef (false)
-part_labels Whether to label individual undef (false)
subparts.
-part_label_merge Whether to merge undef (false)
adjacent subparts when
labeling.
-connector Type of connector to none
use to connect related
features. Options are
"solid," "hat", "dashed",
"quill" and "none".
-all_callbacks Whether to invoke undef
callbacks for autogenerated
"track" and "group" glyphs
-subpart_callbacks Whether to invoke false
callbacks for subparts of
the glyph.
-box_subparts Return boxes around feature 0
subparts rather than around the
feature itself.
-link, -title, -target
These options are used when creating imagemaps
for display on the web. See L</"Creating Imagemaps">.
-filter Select which features to
display. Must be a CODE reference.
Specifying colors: Colors can be expressed in either of two ways:
as symbolic names such as "cyan", as HTML-style #RRGGBB triples, and
r,g,b comma-separated numbers. The symbolic names are the 140 colors
defined in the Netscape/Internet Explorer color cube, and can be
retrieved using the Bio::Graphics::Panel->color_names() method.
Transparent and semi-transparent colors can be specified using the following syntax:
#RRGGBBAA - red, green, blue and alpha
r,g,b,a - red, green, blue, alpha
blue:alpha - symbolic name and alpha
rgb(r,g,b) - CSS style rgb values
rgba(r,g,b,a) - CSS style rgba values
Alpha values can be specified as GD style integers ranging from 0
(opaque) to 127 (transparent), or as CSS-style floating point numbers
ranging from 0.0 (transparent) through 1.0 (opaque). As a special
case, a completely transparent color can be specified using the color
named "transparent". In the rgb() and rgba() forms, red, green, blue
values can be specified as percentages, as in rgb(100%,0%,50%);
otherwise, the values are integers from 0 to 255.
In addition, the -fgcolor and -bgcolor options accept the special
color names "featureScore" and "featureRGB". In the first case,
Bio::Graphics will examine each feature in the track for a defined
"score" tag (or the presence of a score() method) with a numeric value
ranging from 0-1000. It will draw a grayscale color ranging from
lightest (0) to darkest (1000). If the color is named "featureRGB",
then Bio::Graphics will look for a tag named "RGB" and will use that
as the color.
Foreground color: The -fgcolor option controls the foreground color, including the edges of boxes and the like.
Background color: The -bgcolor option controls the background used for filled boxes and other "solid" glyphs. The foreground color controls the color of lines and strings. The -tkcolor argument controls the background color of the entire track.
Track color: The -tkcolor option used to specify the background of the entire track.
Font color: The -fontcolor option controls the color of primary text, such as labels
Secondary Font color: The -font2color option controls the color of secondary text, such as descriptions.
Labels: The -label argument controls whether or not the ID of the feature should be printed next to the feature. It is accepted by all glyphs. By default, the label is printed just above the glyph and left aligned with it.
-label can be a constant string or a code reference. Values can be any of:
-label value Description ------------ -----------
0 Don't draw a label
1 Calculate a label based on primary tag of sequence
"a string" Use "a string" as the label
code ref Invoke the code reference to compute the label
A known bug with this naming scheme is that you can't label a feature with the string "1". To work around this, use "1 " (note the terminal space).
Descriptions: The -description argument controls whether or not a brief description of the feature should be printed next to it. By default, the description is printed just below the glyph and left-aligned with it. A value of 0 will suppress the description. A value of 1 will "magically" look for tags of type "note" or "description" and draw them if found, otherwise the source tag, if any, will be displayed. A code reference will be invoked to calculate the description on the fly. Anything else will be treated as a string and used verbatim.
Connectors: A glyph can contain subglyphs, recursively. The top level glyph is the track, which contains one or more groups, which contain features, which contain subfeatures, and so forth. By default, the "group" glyph draws dotted lines between each of its subglyphs, the "segment" glyph draws a solid line between each of its subglyphs, and the "transcript" and "transcript2" glyphs draw hat-shaped lines between their subglyphs. All other glyphs do not connect their components. You can override this behavior by providing a -connector option, to explicitly set the type of connector. Valid options are:
"hat" an upward-angling conector
"solid" a straight horizontal connector
"quill" a decorated line with small arrows indicating strandedness
(like the UCSC Genome Browser uses)
"dashed" a horizontal dashed line.
The -connector_color option controls the color of the connector, if any.
Collision control: The -bump argument controls what happens when glyphs collide. By default, they will simply overlap (value 0). A -bump value of +1 will cause overlapping glyphs to bump downwards until there is room for them. A -bump value of -1 will cause overlapping glyphs to bump upwards. You may also provide a -bump value of +2 or -2 to activate a very simple type of collision control in which each feature occupies its own line. This is useful for showing dense, nearly-full length features such as similarity hits. Finally, a bump of 3 or the string "fast" will turn on a faster collision-detection algorithm that works well when all features have identical height. The bump argument can also be a code reference; see below.
If you would like to see more horizontal whitespace between features that occupy the same line, you can specify it with the -hbumppad option. Positive values increase the amount of whitespace between features. Negative values decrease the whitespace.
Keys: The -key argument declares that the track is to be shown in a key appended to the bottom of the image. The key contains a picture of a glyph and a label describing what the glyph means. The label is specified in the argument to -key.
box_subparts: Ordinarily, when you invoke the boxes() methods to
retrieve the rectangles surrounding the glyphs (which you need to do
to create clickable imagemaps, for example), the rectangles will
surround the top level features. If you wish for the rectangles to
surround subpieces of the glyph, such as the exons in a transcript,
set box_subparts to a true numeric value. The value you specify will
control the number of levels of subfeatures that the boxes will
descend into. For example, if using the "gene" glyph, set
-box_subparts to 2 to create boxes for the whole gene (level 0), the
mRNAs (level 1) and the exons (level 2).
part_labels: If set to true, each subpart of a multipart feature will be labeled with a number starting with 1 at the 5'-most part. This is useful for counting exons. You can pass a callback to this argument; the part number and the total number of parts will be arguments three and four. For example, to label the exons as "exon 1", "exon 2" and so on:
-part_labels => sub { my ($feature,undef,$partno) = @_; return 'exon '.($partno+1); }
The -label argument must also be true.
part_labels_merge: If true, changes the behavior of -part_labels so that features that abut each other without a gap are treated as a single feature. Useful if you want to count the UTR and CDS segments of an exon as a single unit, and the default for transcript glyphs.
strand_arrow: If set to true, some glyphs will indicate their strandedness, usually by drawing an arrow. For this to work, the Bio::SeqFeature must have a strand of +1 or -1. The glyph will ignore this directive if the underlying feature has a strand of zero or undef.
sort_order: By default, features are drawn with a layout based only on the position of the feature, assuring a maximal "packing" of the glyphs when bumped. In some cases, however, it makes sense to display the glyphs sorted by score or some other comparison, e.g. such that more "important" features are nearer the top of the display, stacked above less important features. The -sort_order option allows a few different built-in values for changing the default sort order (which is by "left" position): "low_score" (or "high_score") will cause features to be sorted from lowest to highest score (or vice versa). "left" (or "default") and "right" values will cause features to be sorted by their position in the sequence. "longest" (or "shortest") will cause the longest (or shortest) features to be sorted first, and "strand" will cause the features to be sorted by strand: "+1" (forward) then "0" (unknown, or NA) then "-1" (reverse).
In all cases, the "left" position will be used to break any ties. To break ties using another field, options may be strung together using a "|" character; e.g. "strand|low_score|right" would cause the features to be sorted first by strand, then score (lowest to highest), then by "right" position in the sequence.
Finally, a subroutine coderef with a $$ prototype can be provided. It
will receive two glyph as arguments and should return -1, 0 or 1
(see Perl's sort() function for more information). For example, to
sort a set of database search hits by bits (stored in the features'
"score" fields), scaled by the log of the alignment length (with
"start" position breaking any ties):
sort_order = sub ($$) { my ($glyph1,$glyph2) = @_; my $a = $glyph1->feature; my $b = $glyph2->feature; ( $b->score/log($b->length) <=> $a->score/log($a->length) ) || ( $a->start <=> $b->start ) }
It is important to remember to use the $$ prototype as shown in the
example. Otherwise Bio::Graphics will quit with an exception. The
arguments are subclasses of Bio::Graphics::Glyph, not the features
themselves. While glyphs implement some, but not all, of the feature
methods, to be safe call the two glyphs' feature() methods in order to
convert them into the actual features.
The '-always_sort' option, if true, will sort features even if bumping is turned off. This is useful if you would like overlapping features to stack in a particular order. Features towards the end of the list will overlay those towards the beginning of the sort order.
-feature_limit: When this option is set to a non-zero value, calls
to a track's add_feature() method will maintain a count of features
added to a track. Once the feature count exceeds the value set in
-feature_limit, additional features will displace existing ones in a
way that effects a uniform sampling of the total feature set. This is
useful to protect against excessively large tracks. The total number
of features added can be retrieved by calling the track's
feature_count() method.
-bump_limit: When bumping is chosen, colliding features will ordinarily move upward or downward without limit. When many features collide, this can lead to excessively high images. You can limit the number of levels that features will bump by providing a numeric bump_limit option. After the limit is hit, features will pile up on top of each other, usually as a band at the bottom of the track.
The -filter option, which must be a CODE reference, will be invoked once for each feature prior to rendering it. The coderef will receive the feature as its single option and should return true if the feature is to be shown and false otherwise.
Instead of providing a constant value to an option, you may subsitute a code reference. This code reference will be called every time the panel needs to configure a glyph. The callback will be called with three arguments like this:
sub callback { my ($feature,$option_name,$part_no,$total_parts,$glyph) = @_; # do something which results in $option_value being set return $option_value; }
The five arguments are $feature, a reference to the IO::SeqFeatureI
object, $option_name, the name of the option to configure,
$part_no, an integer index indicating which subpart of the feature
is being drawn, $total_parts, an integer indicating the total
number of subfeatures in the feature, and finally $glyph, the Glyph
object itself. The latter fields are useful in the case of treating
the first or last subfeature differently, such as using a different
color for the terminal exon of a gene. Usually you will only need to
examine the first argument. This example shows a callback examining
the score() attribute of a feature (possibly a BLAST hit) and return
the color "red" for high-scoring features, and "green" for low-scoring
features:
sub callback { my $feature = shift; if ($feature->score > 90) { return 'red'; else { return 'green'; } }
The callback should return a string indicating the desired value of the option. To tell the panel to use the default value for this option, return the string "*default*".
The callback for -grid is slightly different because at the time this option is needed there is no glyph defined. In this case, the callback will get two arguments: the feature and the panel object:
-glyph => sub { my ($feature,$panel) = @_; return 'gene' if $panel->length < 10_000; return 'box'; }
When you install a callback for a feature that contains subparts, the callback will be invoked first for the top-level feature, and then for each of its subparts (recursively). You should make sure to examine the feature's type to determine whether the option is appropriate.
Also be aware that some options are only called for subfeatures. For
example, when using multi-segmented features, the "bgcolor" and
"fgcolor" options apply to the subfeatures and not to the whole
feature; therefore the corresponding callbacks will only be invoked
for the subfeatures and not for the top-level feature. To get
information that applies to the top-level feature, use the glyph's
parent_feature() method. This returns:
* the parent if called with no arguments or with an argument of (1) * the parent's parent if called with an argument of (2) * the parent's parent's parent if called with an argument of (3) * etc.
The general way to take advantage of this feature is:
sub callback { my ($feature,$option_name,$part_no,$total_parts,$glyph) = @_; my $parent = $glyph->parent_feature();
# do something which results in $option_value being set
return $option_value;
}
or, more concisely:
sub callback { my $feature = shift; # first argument my $glyph = pop; # last argument my $parent = $glyph->parent_feature();
# do something which results in $option_value being set
return $option_value;
}
Some glyphs deliberately disable recursion into subparts. The
"track", "group", "transcript", "transcript2" and "segments" glyphs
selectively disable the -bump, -label and -description options. This
is to avoid, for example, a label being attached to each exon in a
transcript, or the various segments of a gapped alignment bumping each
other. You can override this behavior and force your callback to be
invoked by providing add_track() with a true -all_callbacks
argument. In this case, you must be prepared to handle configuring
options for the "group" and "track" glyphs.
In particular, this means that in order to control the -bump option with a callback, you should specify -all_callbacks=>1, and turn on bumping when the callback is in the track or group glyphs.
The -subpart_callbacks options is similar, except that when this is set to true callbacks are invoked for the main glyph and its subparts. This option only affects the -label and -description options.
The following accessor methods provide access to various attributes of the panel object. Called with no arguments, they each return the current value of the attribute. Called with a single argument, they set the attribute and return its previous value.
Note that in most cases you must change attributes prior to invoking
gd(), png() or boxes(). These three methods all invoke an internal
layout() method which places the tracks and the glyphs within them,
and then caches the result.
Accessor Name Description ------------- -----------
width() Get/set width of panel spacing() Get/set spacing between tracks key_spacing() Get/set spacing between keys length() Get/set length of segment (bp) flip() Get/set coordinate flipping pad_top() Get/set top padding pad_left() Get/set left padding pad_bottom() Get/set bottom padding pad_right() Get/set right padding start() Get the start of the sequence (bp; read only) end() Get the end of the sequence (bp; read only) left() Get the left side of the drawing area (pixels; read only) right() Get the right side of the drawing area (pixels; read only)
The following methods are used internally, but may be useful for those implementing new glyph types.
Return the symbolic names of the colors recognized by the panel object. In a scalar context, returns an array reference.
color_name_to_rgb($color)
Given a symbolic color name, returns the red, green, blue components of the color. In a scalar context, returns an array reference to the rgb triplet. Returns undef for an invalid color name.
rgb($index)
Given a GD color index (between 0 and 140), returns the RGB triplet
corresponding to this index. This method is only useful within a
glyph's draw() routine, after the panel has allocated a GD::Image and
is populating it.
translate_color($color)
Given a color, returns the GD::Image index. The color may be
symbolic, such as "turquoise", or a #RRGGBB triple, as in #F0E0A8.
This method is only useful within a glyph's draw() routine, after the
panel has allocated a GD::Image and is populating it.
set_pen($width,$color)
Changes the width and color of the GD drawing pen to the values
indicated. This is called automatically by the GlyphFactory fgcolor()
method. It returns the GD value gdBrushed, which should be used for
drawing.
You may wish to use Bio::Graphics to create clickable imagemaps for
display on the web. The main method for achieving this is
image_and_map(). Under special circumstances you may instead wish to
call either or both of create_web_image() and create_web_map().
Here is a synopsis of how to use image_and_map() in a CGI script,
using CGI.pm calls to provide the HTML scaffolding:
print h2('My Genome');
my ($url,$map,$mapname) = $panel->image_and_map(-root => '/var/www/html', -url => '/tmpimages', -link => 'http://www.google.com/search?q=$name');
print img({-src=>$url,-usemap=>"#$mapname"});
print $map;
We call image_and_map() with various arguments (described below) to
generate a three element list consisting of the URL at which the image
can be accessed, an HTML fragment containing the clickable imagemap
data, and the name of the map. We print out an <image> tag
that uses the URL of the map as its src attribute and the name of the
map as the value of its usemap attribute. It is important to note
that we must put a "#" in front of the name of the map in order to
indicate that the map can be found in the same document as the
<image> tag. Lastly, we print out the map itself.
image_and_map(@options)
Create the image in a web-accessible directory and return its URL, its clickable imagemap, and the name of the imagemap. The following options are recognized:
Option Description ------ -----------
-url The URL to store the image at.
-root The directory path that should be appended to the
start of -url in order to obtain a physical
directory path.
-link A string pattern or coderef that will be used to
generate the outgoing hypertext links for the imagemap.
-title A string pattern or coderef that will be used to
generate the "title" tags of each element in the imagemap
(these appear as popup hint boxes in certain browsers).
-target A string pattern or coderef that will be used to
generate the window target for each element. This can
be used to pop up a new window when the user clicks on
an element.
-mapname The name to use for the E<lt>mapE<gt> tag. If not provided,
a unique one will be autogenerated for you.
This method returns a three element list consisting of the URL at which the image has been written to, the imagemap HTML, and the name of the map. Usually you will incorporate this information into an HTML document like so:
my ($url,$map,$mapname) = $panel->image_and_map(-link=>'http://www.google.com/search?q=$name'); print qq(<img src="$url" usemap="#$mapname">),"\n"; print $map,"\n";
create_web_image($url,$root)
Create the image, write it into the directory indicated by concatenating $root and $url (i.e. "$root/$url"), and return $url.
create_web_map('mapname',$linkrule,$titlerule,$targetrule)
Create a clickable imagemap named "mapname" using the indicated rules to generate the hypertext links, the element titles, and the window targets for the graphical elements. Return the HTML for the map, including the enclosing <map> tag itself.
To use this method effectively, you will need a web server and an image directory in the document tree that is writable by the web server user. For example, if your web server's document root is located at /var/www/html, you might want to create a directory named "tmpimages" for this purpose:
mkdir /var/www/html/tmpimages chmod 1777 /var/www/html/tmpimages
The 1777 privilege will allow anyone to create files and subdirectories in this directory, but only the owner of the file will be able to delete it.
When you call image_and_map(), you must provide it with two vital pieces of information: the URL of the image directory and the physical location of the web server's document tree. In our example, you would call:
$panel->image_and_map(-root => '/var/www/html',-url=>'/tmpimages');
If you are working with virtual hosts, you might wish to provide the hostname:portnumber part of the URL. This will work just as well:
$panel->image_and_map(-root => '/var/www/html', -url => 'http://myhost.com:8080/tmpimages');
If you do not provide the -root argument, the method will try to figure it out from the DOCUMENT_ROOT environment variable. If you do not provide the -url argument, the method will assume "/tmp".
During execution, the image_and_map() method will generate a unique
name for the image using the Digest::MD5 module. You can get this
module on CPAN and it must be installed in order to use
image_and_map(). The imagename will be a long hexadecimal string such
as "e7457643f12d413f20843d4030c197c6.png". Its URL will be
/tmpimages/e7457643f12d413f20843d4030c197c6.png, and its physical path
will be /var/www/html/tmpimages/e7457643f12d413f20843d4030c197c6.png
In addition to providing directory information, you must also tell
image_and_map() how to create outgoing links for each graphical
feature, and, optionally, how to create the "hover title" (the popup
yellow box displayed by most modern browsers), and the name of the
window or frame to link to when the user clicks on it.
There are three ways to specify the link destination:
By configuring one or more tracks with a -link argument.
By configuring the panel with a -link argument.
By passing a -link argument in the call to image_and_map().
The -link argument can be either a string or a coderef. If you pass a string, it will be interpreted as a URL pattern containing runtime variables. These variables begin with a dollar sign ($), and are replaced at run time with the information relating to the selected annotation. Recognized variables include:
$name The feature's name (display name)
$id The feature's id (eg, PK from a database)
$class The feature's class (group class)
$method The feature's method (same as primary tag)
$source The feature's source
$ref The name of the sequence segment (chromosome, contig)
on which this feature is located
$description The feature's description (notes)
$start The start position of this feature, relative to $ref
$end The end position of this feature, relative to $ref
$length Length of this feature
$segstart The left end of $ref displayed in the detailed view
$segend The right end of $ref displayed in the detailed view
For example, to link each feature to a Google search on the feature's description, use the argument:
-link => 'http://www.google.com/search?q=$description'
Be sure to use single quotes around the pattern, or Perl will attempt
to perform variable interpretation before image_and_map() has a chance
to work on it.
You may also pass a code reference to -link, in which case the code will be called every time a URL needs to be generated for the imagemap. The subroutine will be called with two arguments, the feature and the Bio::Graphics::Panel object, and it should return the URL to link to, or an empty string if a link is not desired. Here is a simple example:
-link => sub { my ($feature,$panel) = @_; my $type = $feature->primary_tag; my $name = $feature->display_name; if ($primary_tag eq 'clone') { return "http://www.google.com/search?q=$name"; } else { return "http://www.yahoo.com/search?p=$name"; }
The -link argument cascades. image_and_map() will first look for a
-link option in the track configuration, and if that's not found, it
will look in the Panel configuration (created during
Bio::Graphics::Panel->new). If no -link configuration option is found
in either location, then image_and_map() will use the value of -link
passed in its argument list, if any.
The -title and -target options behave in a similar manner to -link. -title is used to assign each feature "title" and "alt" attributes. The "title" attribute is used by many browsers to create a popup hints box when the mouse hovers over the feature's glyph for a preset length of time, while the "alt" attribute is used to create navigable menu items for the visually impaired. As with -link, you can set the title by passing either a substitution pattern or a code ref, and the -title option can be set in the track, the panel, or the method call itself in that order of priority.
If not provided, image_and_map() will autogenerate its own title in
the form "<method> <display_name> <seqid>:start..end".
The -target option can be used to specify the window or frame that clicked features will link to. By default, when the user clicks on a feature, the loaded URL will replace the current page. You can modify this by providing -target with the name of a preexisting or new window name in order to create effects like popup windows, multiple frames, popunders and the like. The value of -target follows the same rules as -title and -link, including variable substitution and the use of code refs.
NOTE: Each time you call image_and_map() it will generate a new image
file. Images that are identical to an earlier one will reuse the same
name, but those that are different, even by one pixel, will result in
the generation of a new image. If you have limited disk space, you
might wish to check the images directory periodically and remove those
that have not been accessed recently. The following cron script will
remove image files that haven't been accessed in more than 20 days.
30 2 * * * find /var/www/html/tmpimages -type f -atime +20 -exec rm {} \;
Please report them.
the Bio::Graphics::Glyph manpage, the Bio::Graphics::Glyph::arrow manpage, the Bio::Graphics::Glyph::cds manpage, the Bio::Graphics::Glyph::crossbox manpage, the Bio::Graphics::Glyph::diamond manpage, the Bio::Graphics::Glyph::dna manpage, the Bio::Graphics::Glyph::dot manpage, the Bio::Graphics::Glyph::ellipse manpage, the Bio::Graphics::Glyph::extending_arrow manpage, the Bio::Graphics::Glyph::generic manpage, the Bio::Graphics::Glyph::graded_segments manpage, the Bio::Graphics::Glyph::heterogeneous_segments manpage, the Bio::Graphics::Glyph::line manpage, the Bio::Graphics::Glyph::pinsertion manpage, the Bio::Graphics::Glyph::primers manpage, the Bio::Graphics::Glyph::rndrect manpage, the Bio::Graphics::Glyph::segments manpage, the Bio::Graphics::Glyph::redgreen_box manpage, the Bio::Graphics::Glyph::ruler_arrow manpage, the Bio::Graphics::Glyph::toomany manpage, the Bio::Graphics::Glyph::transcript manpage, the Bio::Graphics::Glyph::transcript2 manpage, the Bio::Graphics::Glyph::translation manpage, the Bio::Graphics::Glyph::triangle manpage, the Bio::Graphics::Glyph::xyplot manpage, the Bio::Graphics::Glyph::whiskerplot manpage, the Bio::SeqI manpage, the Bio::SeqFeatureI manpage, the Bio::Das manpage, GD the GD::SVG manpage glyph_help.pl
Lincoln Stein <lstein@cshl.org>
Copyright (c) 2001 Cold Spring Harbor Laboratory
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself. See DISCLAIMER.txt for disclaimers of warranty.
| Bio::Graphics::Panel - Generate GD images of Bio::Seq objects |