[DAS] Couldn't instantiate datasource ensembl1834_bridge

[Steven Van Vooren]

Dear DAS-list,

First of all, thanks for your comments. Things seem to be working fine now.
I'm now posting some additional questions that have come up.

> Why did compilation fail?  Which version of BioJava were you using?

Using your recent compilation at http://www.derkholm.net/thomasd/das/
(thanks), things are working just fine, so there was no need for me to
compile Dazzle anymore. However, I'm posting this for follow up. After
switching to servlets.jar from apache tomcat 3.3.1a (thanks, Alex), after
grabbing biojava-live from the cvs repository and successfully compiling it,
and after putting everything in the classpath, I still got:

-------- output -------- 
compile:
    [javac] Compiling 20 source files to D:\biodas\xdas\ant-build\classes
    [javac]
D:\biodas\xdas\ant-build\src\org\biojava\servlets\dazzle\datasource\RagbagDa
taSource.java:31: package org.biojava.bio.seq.ragbag does not exist
    [javac] import org.biojava.bio.seq.ragbag.*;
    [javac] ^
[ ... other errors resulting from this ... ]
    [javac]
D:\biodas\xdas\ant-build\src\org\biojava\servlets\dazzle\virtual\VirtualData
Source.java:53: org.biojava.servlets.dazzle.virtual.VirtualDataSource is not
abstract and does not override abstract method
getShatterFeature(org.biojava.bio.seq.Feature) in
org.biojava.servlets.dazzle.datasource.DazzleDataSource
    [javac] class VirtualDataSource implements DazzleDataSource {
    [javac] ^
[ ... other errors resulting from this ... ]
-------- output -------- 

The ragbag package is indeed not present in the biojava-live jar.

> The Dazzle CVS should now be fixed to work with current  CVS
> biojava-live, but it's always possible that I didn't check the changes
> right, or that CVS is doing something strange.

Maybe I'm not using the correct sources. I checked out xdas from the biojava
CVS, I hope this is the correct source for the latest dazzle
(http://cvs.biojava.org/cgi-bin/viewcvs/viewcvs.cgi/xdas/?cvsroot=biojava,
checked out  using cvs -d
:pserver:cvs at cvs.open-bio.org:/home/repository/biojava checkout xdas).

> The skeleton you've downloaded is very old.  You can grab something
> more current from:
>
>        http://www.derkholm.net/thomasd/das/

Thanks for putting this online, it works fine for me. But some new questions
have now come up. Maybe someone on the list can help me out.

[Question 1 - feature storage]

I have set up a MySQL database as described in
http://www.ensembl.org/Docs/das_server_v1.2.pdf (page 4) and configured
dazzle to read annotations from it, which works. In the example described in
that pdf, the contig_id field of the annotations table contains contigs
names (AC012170.6.1.198550). I have tried adding contigs there myself, but
they do not show up when I use the ensembl genome browser to visualise the
region surrounding that contig. Instead, ensembl just reports "no * My...
features in this region".

However, when I put a chromosome number in the contig_id field of that
table, things work fine, and I get the ensembl genome browser to show my
annotations. I assume this is due to the fact that dazzle does not "know"
contig names as entry points (looking at
http://servlet.sanger.ac.uk:8080/das/ensembl1834/entry_points I indeed only
get chromosome numbers). Am I right in this assumption? Is there a way to
enable dazzle to use contig and clone names, instead of putting my
annotations in the database using an offset relative to the "first basepair"
of each chromosome? In the example table in
http://www.ensembl.org/Docs/das_server_v1.2.pdf , they seem to do exactly
that. As the annotations I want to serve for now, coincide perfectly to
clones/contigs, this would save me the conversion step to chromosome
positions.

[Question 2 - links]

>From the LDAS (perl based das server) documentation, it seems to be possible
to serve http-links with annotations. There, a config file tells LDAS how to
construct links from the annotation data. I assume this makes annotations
clickable in the ensembl genome browser. Can I also do this with Dazzle, and
if so, does it require me to add code to the datasource class and compile
them myself, or is there an easy way to serve links? Allowing visitors to
click-through from ensembl to detailed information on the annotations would
be useful in my case.

[Question 3 - speed]

> Both approaches ought to work.  Running your own reference server might
> be faster (and is definitely a good idea if you have your own local
> copy of the ensembl core databases).

In order to run my own reference server, would it suffice just to copy the
homo_sapiens_core_19_34a tables to a local database and point to it in the
config file ( <string name="dbURL"
value="jdbc:mysql://mylocaldbserver/homo_sapiens_core_19_34a" /> ) or are
there other issues to consider?

[Question 4 - caching]

After inserting new annotations to the feature table in MySQL (additionally
to those that were already in the database) the ensembl site only shows
those annotations that were already there. Refreshing the browser window
doesn't solve this. It seems that somewhere, annotations are cached. Is this
done at the ensembl site, or does dazzle cache the annotations itself? How
should can you purge this and force a database-read to include newly added
annotations?

[Question 5 - XML]

1. What is the effect of <boolean name="useHeavyClones" value="true" /> ?

2. I understand that I can use <set name="elideTypes"></set> and <set
name="acceptTypes"></set> to avoid specific feature types to be served. If I
just want dazzle to act as an annotation server with my features, can I use
these settings somehow to speed things up?

[As a reference, this is the XML file I use.]

<dazzle xmlns="http://www.biojava.org/2000/dazzle">

  <resource id="ensembl1934a_bridge" jclass="ensembl.EnsemblHolder">
    <string name="dbURL"
value="jdbc:mysql://ensembldb.ensembl.org/homo_sapiens_core_19_34a" />
    <string name="dbUser" value="anonymous" />
    <string name="dbPass" value="" />
    <boolean name="useHeavyClones" value="true" />

    <int name="schemaVersion" value="19" />
  </resource>

  <datasource id="ensembl1934a" jclass="ensembl.EnsemblReferenceSource">
    <string name="ensemblHolderID" value="ensembl1934a_bridge" />
    <string name="name" value="Ensembl 19.34a" />
    <string name="description" value="The human genome from Ensembl" />
    <string name="version" value="19.34a" />
    <set name="elideTypes">
      <string value="repeat" />
      <string value="similarity" />
      <string value="TSS" />
    </set>
    <string name="stylesheet" value="ensemblbase-style.xml" />
    <string name="ensemblWebURL"
value="http://www.ensembl.org/Homo_sapiens/" />
  </datasource>

<!-- My annotations from mySQL -->
<datasource id="MyDataSource"
jclass="ensembl.EnsemblGenericSeqFeatureSource">
<string name="ensemblHolderID" value="ensembl1934a_bridge" />
<string name="name" value="Ensembl Test Source" />
<string name="description" value="My DAS feature source" />
<string name="version" value="1.00" />
<string name="mapMaster"
value="http://servlet.sanger.ac.uk:8080/das/ensembl1934a" />
<string name="dbURL" value="jdbc:mysql://mysqlhost/ensembl_generic" />
<string name="dbUser" value="****" />
<string name="dbPass" value="****" />
<string name="tableName" value="my_feature" />
<string name="stylesheet" value="generic-style.xml" />
</datasource>

</dazzle>

Thank you,
Steven