M-CHIPS Storage Concept

Kurt Fellenberg

Abstract

The M-CHIPS Database concept allows for data analysis of all of its components including the experimental annotations. It accounts for rapid increase in amount of hybridisation data, experimental descriptions becoming more detailed and new kinds of experiments we expect for the future. This report aims to present our work on storing microarray data from the more general aspects down to the practical details.

Definitions

Relations, attributes and tuples

A relational database consists of

• relations, also called classes or tables. Such a table relates between
• attributes also referred to as data fields or columns of such a table and may contain an arbitrary number of
• tuples, also called records or datasets, which are represented as the rows of a table.

Query

A database query retrieves a set of tuples from a specified relation which meet a certain condition (also specified in the query).

Indices

The purpose of an index is to allow rapid access to specified values within a relation. Without an index, the server process executing a query has to read from the beginning of the table to the end, looking for relevant tuples. An index is computed on a relation for one (or several) of its attributes, ordering the comprised values and storing pointers into the relation in an appropriate data structure (e.g. a b-tree or hash table). A query is directed to the matching rows by the index, thus performing much faster than without.

Inheritance

The PostgreSQL database management system (DBMS) allows for construction of table hierarchies by inheritance. Let the table B be created inheriting from table A, then

• B inherits all attributes of A and
• its tuples are accessible by querying the parental table A
• by anyone having read permission on A regardless of permissions on B.

Data integrity

Suppose a valid alteration of the data, defined by a block of sequentially permformed operations (such a block is called a 'transaction') breaks down after doing half of the work. Maybe a table has been deleted but remains existant in the table administrating system catalogue of the database system. Or the task was to add $500 to everyone's salary in a table containing employees and now it is unknown which row was updated and which not. In both cases data integrity (database consistency) is violated.

 

• Abstract
• Definitions
  • Relations, attributes and tuples
  • Query
  • Indices
  • Inheritance
  • Data integrity
• Contents
• 1. General Considerations concerning Data Storage
  • 1.1 Storing transcription intensities
    • 1.1.1 What to store
    • 1.1.2 How to store it
  • 1.2 Storing gene annotations
    • 1.2.1 What to store
    • 1.2.2 How to store it
  • 1.3 Storing experimental annotations
    • 1.3.1 What to store
    • 1.3.2 How to store it
      • 1.3.2.1 Flexibility
      • 1.3.2.2 Performance
      • 1.3.2.3 Analysis aspects
      • 1.3.2.4 M-CHIPS storage
• 2. Realization concepts
  • 2.1 Safety considerations
    • 2.1.1 Global accidents
      • 2.1.1.1 Transactions
      • 2.1.1.2 Global backups
    • 2.1.2 Private errors
    • 2.1.3 Unauthorized access
  • 2.2 Performance considerations
    • 2.2.1 Minimizing query space
    • 2.2.2 Separate or block-wise storage
  • 2.3 Flexibility considerations
  • 2.4 Considerations related with Analysis
• 3. Implementation
  • 3.1 Databases
  • 3.2 Microarray Families
    • 3.2.1 Gene annotations
    • 3.2.2 Administration of the comprised multiconditional experiments
  • 3.3 Multiconditional Experiments
    • 3.3.1 Experimental annotations constant within the multiconditional experiment
    • 3.3.2 Administration of the comprised condiditions
  • 3.4 Experimental Conditions
    • 3.4.1 Administration of the comprised hybridizations and measurements
    • 3.4.2 Experimental annotations dependent on the condition
  • 3.5 Hybridisations
    • 3.5.1 Hybridisation intensities
    • 3.5.2 Hybridisation intensities, `solidified'
    • 3.5.3 Measurement-dependent experimental annotations