Search extension transforms Wiki into a relational system: A case for flavonoid metabolite database

Masanori Arita,Kazuhiro Suwa

doi:10.1186/1756-0381-1-7

Abstract

BackgroundIn computer science, database systems are based on the relational model founded by Edgar Codd in 1970. On the other hand, in the area of biology the word 'database' often refers to loosely formatted, very large text files. Although such bio-databases may describe conflicts or ambiguities (e.g. a protein pair do and do not interact, or unknown parameters) in a positive sense, the flexibility of the data format sacrifices a systematic query mechanism equivalent to the widely used SQL.ResultsTo overcome this disadvantage, we propose embeddable string-search commands on a Wiki-based system and designed a half-formatted database. As proof of principle, a database of flavonoid with 6902 molecular structures from over 1687 plant species was implemented on MediaWiki, the background system of Wikipedia. Registered users can describe any information in an arbitrary format. Structured part is subject to text-string searches to realize relational operations. The system was written in PHP language as the extension of MediaWiki. All modifications are open-source and publicly available.ConclusionThis scheme benefits from both the free-formatted Wiki style and the concise and structured relational-database style. MediaWiki supports multi-user environments for document management, and the cost for database maintenance is alleviated.

Highlights

Introduction of fully dependent pagesAs a natural extension of page dependency, we can let the system generate fully dependent pages
Information tables for compounds, plant species, and references Our basic concept is to describe tabulated data in Wiki pages and to let all other operations such as formatting for visualization, obtaining statistics, or applying relational operators be done by the template mechanism and textstring searches
Note that search results are labeled with page titles, which serve as the identifier of the relation table being searched

Summary

Introduction

Introduction of fully dependent pagesAs a natural extension of page dependency, we can let the system generate fully dependent pages. We prepared two types of fully dependent pages: those that are not necessarily saved, and those that should be saved and subject to page (or internet) searches. Examples of the former are different display styles of an identical page or volatile information that needs not be searched for. Examples of the latter are the index pages. Management, and query are the keys How can we promote the development and maintenance of high-quality databases? Still unsupported is flexibility in query mechanisms and presentation such as displaying customized statistics in a user-friendly way

Methods

Results

Discussion

Conclusion