MEGGASENSE - The Metagenome/Genome Annotated Sequence Natural Language Search Engine: A Platform for the Construction of Sequence Data Warehouses.

Ranko Gaćeša ,Paul F Long,Daslav Hranueli,John Cullum,Janko Diminić ,Ólafur H Friðjónsson ,Sólveig K Pétursdóttir ,Guðmundur Ó Hreggviðsson ,Elísabet Eik Guðmundsdóttir ,Jurica Žučko ,Antonio Starčević

doi:10.17113/ftb.55.02.17.4749

Abstract

The MEGGASENSE platform constructs relational databases of DNA or protein sequences. The default functional analysis uses 14 106 hidden Markov model (HMM) profiles based on sequences in the KEGG database. The Solr search engine allows sophisticated queries and a BLAST search function is also incorporated. These standard capabilities were used to generate the SCATT database from the predicted proteome of Streptomyces cattleya. The implementation of a specialised metagenome database (AMYLOMICS) for bioprospecting of carbohydrate-modifying enzymes is described. In addition to standard assembly of reads, a novel 'functional' assembly was developed, in which screening of reads with the HMM profiles occurs before the assembly. The AMYLOMICS database incorporates additional HMM profiles for carbohydrate-modifying enzymes and it is illustrated how the combination of HMM and BLAST analyses helps identify interesting genes. A variety of different proteome and metagenome databases have been generated by MEGGASENSE.

Highlights

Falling costs of generation sequencing have made de novo genome and metagenome sequencing widely avail able
The default functional analysis for databases generated by MEGGASENSE is derived from the KEGG database
The analyses described above are incorporated as a default in metagenome databases generated by MEGGASENSE

Summary

Introduction

Falling costs of generation sequencing have made de novo genome and metagenome sequencing widely avail able. Bioinformatics offers many tools to analyse the sequences, and the identification of protein-coding regions and assignment of function are the major aim in most projects. There are many tools to try to assign function to such proteins. A general BLAST database such as GenBank (3) consists mainly of uncurated entries, which will often contain misleading data for functional assignment. The SEED database (4) contains collections of protein sequences grouped by function and has been used for BLAST searches to find hits corresponding to in silico translation of the metagenomic sequences. In order to present functional information about the whole genome or metagenome effectively, it is necessary to have a suitable data structure. BLAST searches against the KEGG orthologues are a useful way of assigning function to new sequences

Methods

Results

Conclusion