Skip to content
Snippets Groups Projects
Commit f092ffe8 authored by edgrif's avatar edgrif
Browse files

first versions

parent 7584342a
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
web/user_doc/Cluster_tags.shtml 0 → 100755
+ 174
− 0
View file @ f092ffe8
<!--#set var="banner" value="ZMap Feature Sets and Styles"-->
<!--#include virtual="/perl/header"-->
<!--#set var="author" value="edgrif@sanger.ac.uk" -->
<style>
pre{ width: 95%; background-color: #DDDDDD; border-style: solid; border-width: 1px; padding: 10px }
.example{ border-color: #000000 }
</style>
<br/>
<fieldset>
<legend id="Cluster_tags">Clustering Features In Acedb</legend>
<h3>Introduction</h3>
<p>Please see the <a href="./Tag_sets.shtml">Tag Set</a> documentation for a key to
colouring in Acedb tag examples.</p>
<p>For a number of reasons we wish to cluster together features for processing in various
ways. Obvious examples are:
<ul>
<li><p>showing matching EST read pairs.</p>
<li><p>showing which HSP's were connected by the alignment program used to find them.</p>
<li><p>showing/processing as a unit the transcript and the matches that align to the
exons of that transcript.</p>
</ul>
<p>Acedb has provided two main ways of clustering:</p>
<ul>
<li><p><b>Explicit:</b> using tags in objects to "point" to other sibling/parent/child
objects, these usually make use of the XREF mechanism to automatically maintain lists
of these relationships.</p>
<li><p><b>Implicit:</b> Certain tags, e.g. Homols, EST_n are clustered by the code for certain
types of display.</p>
</ul>
<p>With more complex analysis of genetic data a more generalised way of clustering features is
required. We need a mechanism that can specify different kinds of clustering and which can be
used for clustering both acedb objects (e.g. transcripts) and features within those
objects (e.g. homologies) and any combination of those types.</p>
<p>The method for clustering must be able to represent different types of clustering using a single
generic "cluster" object. This document describes a set of tags for achieving this.</p>
<h3>The Acedb Model for EST Read Pairs</h3>
<p>Acedb has several tags in the ?Sequence class to implement clustering of EST read pairs:</p>
<pre class="model_clases">
?Sequence
Visible
Paired_read ?Sequence XREF Paired_read
Properties
cDNA
EST_5
EST_3
Show_in_reverse_orientation
</pre>
<p>The meaning of these tags is:</p>
<ul>
<li><b>Paired_read:</b> used to cross reference to the matching read sequence object (arguably
the tag should be followed by UNIQUE to ensure there were only two objects).
<li><b>EST_5 or EST_3:</b> indicates which object is 5' and which 3'
<li><b>Show_in_reverse_orientation:</b> Usually the 3' read is sequenced in reverse, this
flag signals that the object should be reverse complemented for display.
</ul>
<p>This system is tailored to EST read pairs but we can adapt it to a more general model as detailed
in the following section.</p>
<h3>Tag Sets for Clustering</h3>
<p>The new tags will use some of the existing acedb tags, modify others and introduce
some new ones.</p>
<h4>EST tags</h4>
<pre class="model_clases">
?Sequence
Properties
cDNA
EST_5
EST_3
</pre>
<p>There are two choices with these tags:</p>
<ol>
<li><b>Make them acedb "Special" tags:</b> Currently acedb code checks that read pairs are comprised of
EST_5 and EST_3 pairs, not necessarily a good thing.....
<li><b>Use them as normal tags:</b> remove the "specialness" of these tags and have them simply as information
for the annotator, i.e. the code does not check them at all.
</ol>
<p>If they are to be special tags then it could be argued that they should be embedded
in the Cluster_pair tags as they would have semantics only in that context. If they are not
special tags then they do not need to be changed.</p>
<p>I propose they stay asis.</p>
<h4>Show_in_reverse_orientation tag</h4>
<pre class="model_clases">
?Sequence
Properties
Show_in_reverse_orientation
</pre>
<p>This tag is a separate issue from the Cluster tag but is included because it has an
important role in the display of read pairs in that it is usual that the 3' end is
sequenced in reverse and so needs to be reverse complemented before display.
We may find there are other display types we would like to apply to Clustered features
in which case we could augment this with other tags.</p>
<p>I propose this tag stays asis.</p>
<h4>Cluster tags</h4>
<p>In Acedb the Read_pair tag was used to cluster pairs of sequence objects, this should
be replaced with a more general model, the following examples use the "Tag2" convention
to allow arbitrary addition of different kinds of object to the tag set.
<p>To cluster pairs of objects:</p>
<pre class="model_classes">
<font color=red>Cluster_Pair</font> <font color=green>Seq_pair</font> <font color=red>UNIQUE</font> <font color=green>?Sequence</font> <font color=red>XREF</font> <font color=green>Seq_pair</font>
<font color=green>XX_type</font> <font color=red>UNIQUE</font> <font color=green>?XXX</font> <font color=red>XREF</font> <font color=green>XXX_type</font>
</pre>
<p>This definition ensures that objects are paired.</p>
<p>To cluster objects into sets/trees:</p>
<pre class="model_classes">
<font color=red>Cluster_Tree Parent UNIQUE</font> <font color=green>Seq_parent</font> <font color=red>UNIQUE</font> <font color=green>?Sequence</font> <font color=red>XREF</font> <font color=green>Seq_children</font>
<font color=green>XXX_parent</font> <font color=red>UNIQUE</font> <font color=green>?XXX</font> <font color=red>XREF</font> <font color=green>XXX_children</font>
<font color=red>Children</font> <font color=green>Seq_children ?Sequence</font> <font color=red>XREF</font> <font color=green>Seq_parent</font>
<font color=green>XXX_children ?XXX</font> <font color=red>XREF</font> <font color=green>XXX_parent</font>
</pre>
<p>The tags impose the following rules:</p>
<ul>
<li><p>There can be an arbitrary depth of child/parent clustering.</p>
<li><p>The child/parent tree is a DAG with single parents only.</p>
</ul>
</fieldset>
<!--#include virtual="/perl/footer"-->
web/user_doc/Tag_sets.shtml 0 → 100755
+ 54
− 0
View file @ f092ffe8
<!--#set var="banner" value="ZMap Feature Sets and Styles"-->
<!--#include virtual="/perl/header"-->
<!--#set var="author" value="edgrif@sanger.ac.uk" -->
<style>
pre{ width: 95%; background-color: #DDDDDD; border-style: solid; border-width: 1px; padding: 10px }
.example{ border-color: #000000 }
</style>
<br/>
<fieldset>
<legend id="Tag_sets">Tag Sets In Acedb</legend>
<H3>Tag Sets</H3>
<P>A "tag set" is a set of tags and data that occur in a defined order and can be processed
by acedb code regardless of the class they appear in.
These tag sets are colour coded in this document to help
identify the significant parts of the tag set:
<P><pre><code><font color=red>feature_tag</font> <font color=green>[anonymous tag and object reference]</font> <font color=purple>[feature specific tags and data]</font>
</pre></code>
<P>Where:
<P><font color=red>feature_tag</font> is the tag that the code searches for and locates
on to find out what sort of feature it is processing. This tag must be specified <b>exactly</b>
as given in these examples.
<P><font color=green>anonymous tag and object reference</font> are sometimes included
to allow insertion in to the tag set of object references of arbitrary class (this is also known
as the "tag2 system"). Although this anonymous tag must be present, it's value is not read
by the code and so it can have any value. Similarly while the anonymous object reference
must be present, the class of the object is not used by the code and so it can be any class.</p>
<P><font color=purple>feature specific tags and data</font> are specific tags and data that follow
a particular feature_tag and must be in the order and of the type specified in the tag set description.</p>
<P>Some examples:
<P><pre><code><font color=red>Source_Exons</font><font color=purple> Int UNIQUE Int</font>
<font color=red>Homol</font> <font color=green>DNA_homol ?Sequence</font> <font color=red>XREF</font> <font color=green>DNA_homol</font> <font color=purple>?Method Float Int Int Int Int #Homol_info</font>
</pre></code>
</fieldset>
<!--#include virtual="/perl/footer"-->
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment