GERMINATE A Plant Data Management System

1
GERMINATEA Plant Data Management System

• Jennifer Lee (UoD)
• Guy Davenport (JIC)
• Andy Flavell (UoD)
• Dave Marshall (SCRI)
• Robbie Waugh (SCRI)
• Jo Dicks (JIC)
• Noel Ellis (JIC)
• Mike Ambrose (JIC)
• Theo van Hintum (CGN)

2
GERMINATE Project

• http//bioinf.scri.sari.ac.uk/germinate/
• Developed in PostgreSQL
• Expected to be released under the GNU public
  license open source
• Core of the system is based on FAO/IPGRI multi
  crop passport descriptors
• Additional generic tables to accommodate other
  data types.

3
GERMINATE Project
4
The Database

• Designed to potentially hold any type of data
  associated with plants.
• Accession is the entry level to all data. This
  permits complex queries using vastly different
  dataset.
• Comments related to any item in the database can
  be stored and accessed easily.

5
PK Primary Key FK Foreign Key U Unique
Index I Index
6
PK Primary Key FK Foreign Key U Unique
Index I Index
7
Loading Data

• Most data can be loaded directly from Excel
  spreadsheets or MS Access tables.
• The data is either copied directly into the
  database (MS Access tables) or modified by a Perl
  script (for Excel spreadsheets) from a standard
  format. The data is then moved into the
  appropriate tables.

8
(No Transcript)
9
Passport Data
10
Passport Data
11
PK Primary Key FK Foreign Key U Unique
Index I Index
12
Some Curation Necessary
Taxonomy example
13
Collecting Example
Example Accession CGN03335
AccessionCollecting
Collecting
CollectingSites
Collecting and CollectingSites are not updated.
AccessionCollecting is if accession_id and
collecting_id match.
14
Curation

• Pre-processing interface to allow users to see
  what information is in the database before they
  submit their data.
• Web based curation tool.
• Developed at Iowa State University
• Submitted data must be validated before it is
  available to the public.

15
Data Interaction
PK Primary Key FK Foreign Key U Unique
Index I Index
16
Data Association Levels
PK Primary Key FK Foreign Key U Unique
Index I Index
17
Genetic Data

• Data stored as 2D array
• Accessions as dimension 0
• Markers as dimension 1
• Data is stored as an integer and decoded in the
  EnumUnits and EnumUnitsArrays table
• We are using an Allele Index approach which will
  simplify queries significantly in cases where the
  marker type is not important, but only the
  relative allele values are required.
• Experiment information includes author, date, an
  experiment name and brief description.
• Method information includes a name and
  description (In general more general than
  experiment, multiple experiments are expected to
  use the same method.) and a link to the units.
• Units indicate the marker type.

18
Database
Genetic Data in GERMINATE
metadataset 2
metadataset 3
dataset 1
dataset 1
dataset 1
dimension1
dimension0
data (integer data)
Original Data
accessions (reference data)
markers (string data)
integer_data (enum_index)
dataset_id
index_id
string_data
dataset_id
index_id
reference_id
dataset_id
index_id
table_id
5 -gt Accessions table reference_id accession_id
Accessions
Datasets
Metadatasets
accession_id
germinate_id
instcode
accenumb
experiment_id
metadataset_id
data_type_id
dataset_id
method_id
dimension
dataset_id
dataset_discription
size
19
Decoding the Allele Index
enum_table_array_text
Relative comparison
enum_table_array_int
allele_index
enum_index
allele_value
text
unit_id
enum_index
allele_value
enum_index
enum_value
int
array
unit 7 data_value
unit 8 data_value
unit 9 data_value
unit_id
enum_index
enum_value
array

• Allows expansion of allele values to any ploidy
  level.
• Match of enum_index between unit types do not
  indicate match in allele phase.
• Much more efficient than storing the allele
  values as allele, while still permitting
  searching of individual alleles in any polyploidy
  level.
• If marker type is not important in query the
  methods, units and enum tables can be bypassed in
  a join query to speed up queries.

20
Genetic Map Data

• 3 sets of data
• Population
• Stored in Pedigree table, reference to
  individuals in reference table which links
  population to the dataset.
• Data used to create map
• Stored similar to genetic data
• Map

21
Genetic Map Data
Original Data
metadataset 2
metadataset 3
dataset 1
dataset 1
dataset 1
dimension0
dimension1
loci
position
linkage groups
string data
real data
string data
string_data
dataset_id
dataset_id
real_data
dataset_id
index_id
index_id
string_id
index_id
Additional information can be stored as added
dimensions to the dataset.
22
Image data

• Store either images or links to images for access
  from an interface.
• We should be able to map images
• For example in the case of microarray images the
  spots can be mapped to the accession/sample and
  bring up information on that accession.

23
(No Transcript)
24
Interface

• Currently we have a light weight Perl-CGI
  interface
• Working towards a more flexible interface that
  would allow complex query formation from users.
• Return results as objects will allow navigation
  through data without searching again.

25
Acknowledgments

• University of Dundee
• Andy Flavell
• Scottish Crop Research Institute (SCRI)
• David Marshall
• Robbie Waugh
• Centre for Genetic Resources, The Netherlands
  (CGN)
• Theo van Hintum

• John Innes Centre (JIC)
• Guy Davenport
• Jo Dicks
• Mike Ambrose
• Noel Ellis
• Funding
• BBSRC Grant 94/BEP17084, the Bioinformatics and
  E-science program