The MethDB DAS Server

1
The MethDB DAS Server

• Adding an Epigenetic Information Layer to the
  Human Genome
• Vincent Negre
• Christoph Grunau

2
Overview

• Introduction
• How does it work? ?DAS
• Examples

3
Introduction

• The database is the major source for
• experimentally confirmed DNA methylation data.
• The database is designed to store and annotate
  information about the occurrence of methylated
  cytosines in the DNA.
• Recently, MethDB provided the entire training
  data set for the development of an algorithm to
  predict methylated sites in the human genome
  (Bhasin M et al Prediction of methylated CpGs in
  DNA sequences using a support vector machine).
• The database contains now 19905 methylation
  content data and 5382 methylation patterns or
  profiles for 48 species, 1511 individuals, 198
  tissues and cell lines and 79 phenotypes.

4
Introduction

• Searching
• In the first version of MethDB the Data was
  gathered by annotators, and the database could
  only be queried.
• In the second stage, they added an online data
  submission system that is open to the public.
• Integration of MethDB into the network of
  biological databases via the Distributed
  Annotation System (DAS).
• Thanks to the DAS it is possible to see the
  epigenetic data as a layer to the human Genome.

5
Motivation

• what defines the genomic regions that become
  unmethylated or methylated ?
• still not entirely clear
• maybe the position of different information like
  sites of transcription, exon positions and
  epigenetic data will help us to know the reasons

6
How does it work?
7
distributed annotation system

• The Distributed Annotation System (DAS) defines
  a communication protocol used to exchange
  biological sequence annotation.
• Motivation Annotations should not be provided
  by single centralized databases, but should
  instead be spread over multiple sites.
• It allows a single machine to gather up sequence
  annotation information from multiple distant web
  sites, collate the information, and display it to
  the user in a single view.

8
distributed annotation system
Illustration
9
distributed annotation system

• Annotation Server for MethDB is based on the
  Lightweight DAS (LDAS) server package
  (http//www.biodas.org/servers/)
• Annotations are anchored to a reference sequence
• Human DNA sequences for which methylation data
  are available in MethDB were aligned to the
  Ensembl reference sequence, and LDAS compatible
  annotation files were generated with
  eL-DASionator

10
MethDBEnsembl

• Ensembl is a popular genome browser that uses the
  DAS protocols.
• The DAS of MethDB uses Ensembl (www.ensembl.org)
  as a reference server.
• Alignation of MethDB Sequences via LDAS with
  Ensembl reference Sequences.

11
MethDBEnsembl

• Binding the two Databases Step by Step (we will
  do it now together). This procedure has to be
  done only once.
• After attachment ? regions of methylation data
  represented as colored rectangles.
• The Annotations are hyperlinke to MethDB.
• MethDB is the only DNA methylation database that
  directly integrated into Ensembl.

12
MethDBEnsembl

• To show the abilities of the MethDB DAS ?
  downloaded 31312 Seq. in FASTA format from the
  CpG island tagging project. (www.sanger.ac.uk/HGP/
  cgi.shtml).
• In this form, the sequences could not be queried,
  and no information about the location on the
  genome was present.
• Filtering assambling into 13786 contigs that
  experimentally confirmed as hypomethylated areas
  in the genome.

13
Examples

• Cross conformation of BRCA1 - MethDB and
  predicted CGI
• The MethDB annotation confirms the existence of
  an experimentally confirmed hypomethylated area
  in this region.
• All independent data sources point toward the
  same area.

14
Examples

• CALCAReconstruction of a CGI (Calcitonin to
  reduce blood calcium)
• Prediction depends naturally on the chosen
  parameters, and as any prediction needs
  experimental confirmation.
• Because of technical reasons CGI can not be
  covered entirely but must be reconstructed.
• Combination of bioinformatic tools can predict
  the most probable CGI.

15
Examples

• GSTP1Generation of methylation profiles along a
  region (play an important role in detoxification
  )
• Technical constraints limit the length of
  sequences that can analyzed for site-specific
  methylation.
• Solution ? analysis of overlapping or neighboring
  sequence fragments.
• As a epigenetic Layer we can see where the CGI
  spans.

16
Examples

• Experiment planning Identification of new CGIs
• Predicted CGI often span several hundred base
  pairs (sometimes 3kb), and a detailed analysis of
  the entire region is costly and labor-intensive.
• MethDB can show experimental proofed
  hypomethylated sub regions within the predicted
  CGI.
• With this information better experimental
  planning can be done.