Taxonomy-Dependent and Independent Binning Methods in Metagenomics

Organizing Metagenomic Data: Classification Techniques

A typical workflow in metagenomics generates an array of sequencing reads, contigs, and identified genes. Linking these elements to the original organisms they came from is crucial for understanding the ecological makeup. This task of associating sequences with originating organisms or broader taxonomic groups is commonly referred to as binning or classification.

In the context of shotgun sequencing, data analysis aims to detail both the taxonomic and functional diversity within a specific environment by evaluating DNA fragments from the native microbial population. Binning methods fall into two broad categories: those dependent on known taxonomies and those that aren’t.

Taxonomy-Dependent Methods:

Most existing techniques for sorting shotgun-sequencing data rely on known taxonomies. These approaches gauge ‘similarity’ between sequence reads and either existing sequences in databases or models created from those databases. Depending on the strategy for assessing this similarity, taxonomy-dependent techniques can be further divided into alignment-based, composition-based, and hybrid models. In our research, we introduced MBMC (Metagenomic Binning by Markov Chains), a unique taxonomy-dependent method that doesn’t require alignment.

Taxonomy-Independent Methods:

These techniques cluster sequence reads based on inherent similarities, without the need for database comparisons.

Unsupervised models in this category usually sort reads based on a few key observations:
– The frequency of k-mers within a genome’s reads often directly correlates to the abundance of that genome.
– Distinctively long w-mers are generally unique to each genome.
– The distribution of short q-mers in individual, sufficiently long reads from the same or similar genomes tends to be comparable.

In this regard, we’ve developed a taxonomy-independent technique known as MBBC (Metagenomic Binning Based on Composition).

The following table shows a list of commonly used tools for metagenomic binning.

There is no standard for the taxonomic classification of metagenome sequences. Also, taxonomic sequence classification can be error prone, in particular for habitats with a complex diversity or high proportions of as yet barely characterized taxa.