Spherical: an iterative workflow for assembling metagenomic datasets

Thomas C A Hitch,Christopher J Creevey

doi:10.1186/s12859-018-2028-2

Abstract

BackgroundThe consensus emerging from the study of microbiomes is that they are far more complex than previously thought, requiring better assemblies and increasingly deeper sequencing. However, current metagenomic assembly techniques regularly fail to incorporate all, or even the majority in some cases, of the sequence information generated for many microbiomes, negating this effort. This can especially bias the information gathered and the perceived importance of the minor taxa in a microbiome.ResultsWe propose a simple but effective approach, implemented in Python, to address this problem. Based on an iterative methodology, our workflow (called Spherical) carries out successive rounds of assemblies with the sequencing reads not yet utilised. This approach also allows the user to reduce the resources required for very large datasets, by assembling random subsets of the whole in a “divide and conquer” manner.ConclusionsWe demonstrate the accuracy of Spherical using simulated data based on completely sequenced genomes and the effectiveness of the workflow at retrieving lost information for taxa in three published metagenomics studies of varying sizes. Our results show that Spherical increased the amount of reads utilized in the assembly by up to 109% compared to the base assembly. The additional contigs assembled by the Spherical workflow resulted in a significant (P < 0.05) changes in the predicted taxonomic profile of all datasets analysed. Spherical is implemented in Python 2.7 and freely available for use under the MIT license. Source code and documentation is hosted publically at: https://github.com/thh32/Spherical.

Highlights

The consensus emerging from the study of microbiomes is that they are far more complex than previously thought, requiring better assemblies and increasingly deeper sequencing
Quality analysis of resulting assemblies We used a simulated metagenomic dataset [22] created from 400 species of varying abundance to investigate the accuracy of contigs produced by the Spherical workflow
The secondary iterations carried out allowed alignment of an additional 5.25% of the raw reads compared to the base assembly

Summary

Introduction

The consensus emerging from the study of microbiomes is that they are far more complex than previously thought, requiring better assemblies and increasingly deeper sequencing. Current metagenomic assembly techniques regularly fail to incorporate all, or even the majority in some cases, of the sequence information generated for many microbiomes, negating this effort This can especially bias the information gathered and the perceived importance of the minor taxa in a microbiome. By far the biggest issue with metagenomic sequencing datasets is the resulting uneven coverage of the taxa from the microbiome arising from the complexity and uneven distribution of species in natural microbial communities [4]. This leads to over-sequencing of dominant species in the community and heavily fragmented assemblies of the genomes of minority species, if they can be assembled at all [13].

Methods

Results

Discussion

Conclusion