Unsupervised statistical clustering of environmental shotgun sequences

Andrey Kislyuk,Joshua S Weitz,Srijak Bhatnagar,Jonathan Dushoff

doi:10.1186/1471-2105-10-316

Abstract

BackgroundThe development of effective environmental shotgun sequence binning methods remains an ongoing challenge in algorithmic analysis of metagenomic data. While previous methods have focused primarily on supervised learning involving extrinsic data, a first-principles statistical model combined with a self-training fitting method has not yet been developed.ResultsWe derive an unsupervised, maximum-likelihood formalism for clustering short sequences by their taxonomic origin on the basis of their k-mer distributions. The formalism is implemented using a Markov Chain Monte Carlo approach in a k-mer feature space. We introduce a space transformation that reduces the dimensionality of the feature space and a genomic fragment divergence measure that strongly correlates with the method's performance. Pairwise analysis of over 1000 completely sequenced genomes reveals that the vast majority of genomes have sufficient genomic fragment divergence to be amenable for binning using the present formalism. Using a high-performance implementation, the binner is able to classify fragments as short as 400 nt with accuracy over 90% in simulations of low-complexity communities of 2 to 10 species, given sufficient genomic fragment divergence. The method is available as an open source package called LikelyBin.ConclusionAn unsupervised binning method based on statistical signatures of short environmental sequences is a viable stand-alone binning method for low complexity samples. For medium and high complexity samples, we discuss the possibility of combining the current method with other methods as part of an iterative process to enhance the resolving power of sorting reads into taxonomic and/or functional bins.

Highlights

The development of effective environmental shotgun sequence binning methods remains an ongoing challenge in algorithmic analysis of metagenomic data
The field is driven by environmental shotgun sequencing (ESS), a technique of applying highthroughput genome sequencing to non-clonal DNA purified directly from an environmental sample
We discuss numerical methods used in the implementation, including a novel coordinate transformation which achieves dimension reduction for the feature space of k-mer frequencies, and the genomic fragment divergence measure Dn, a novel statistical measure we developed for performance evaluation of our algorithm

Summary

Introduction

The development of effective environmental shotgun sequence binning methods remains an ongoing challenge in algorithmic analysis of metagenomic data. The field is driven by environmental shotgun sequencing (ESS), a technique of applying highthroughput genome sequencing to non-clonal DNA purified directly from an environmental sample. This removes the requirement to isolate and cultivate clonal cultures of each species, allowing an unprecedented broad view of microbial communities. Attention has been directed to bacterial and viral fractions of these communities, with eukaryotic metagenomics pioneered by projects such as the marine protist census [7]. Complexity of these communities varies greatly from 5 to several thousand identifiable bacterial species. "deep sequencing" using 454 pyrosequencing suggests that possibly tens of thousands of species coexist in a single ml of seawater [10]

Methods

Results

Conclusion