Natrix2 brings Nanopore reads into an open-source metabarcoding workflow

Oxford Nanopore Technologies sequencing can generate exceptionally long DNA reads, offering researchers a way to examine complete amplicons and explore biodiversity in environmental samples with fewer fragmented sequences. Yet those reads also carry higher error rates than the short reads produced by Illumina platforms, and turning them into reliable taxonomic observations requires a carefully coordinated series of trimming, orientation, clustering and consensus-generation steps. A new version of the open-source Natrix workflow aims to make that process more accessible by adding dedicated Nanopore support, a second clustering method, new taxonomic classifiers and reference databases tailored to fungal and eukaryotic research.

The upgrade, presented in Metabarcoding and Metagenomics, builds on Natrix, a pipeline designed for processing amplicon data from environmental samples. Marker regions such as the 16S and 18S ribosomal RNA genes are commonly sequenced to determine which organisms are present and to estimate the relative abundance of their reads in a microbial or eukaryotic community. Natrix originally combined quality control, read assembly, dereplication, chimera detection, clustering and taxonomic assignment in a configurable workflow built with Snakemake and the Bioconda package manager. That integrated design can reduce the complexity of chaining together separate software tools, while version-controlled dependencies support reproducibility and scalable analysis.

The original Illumina-oriented workflow included tools for demultiplexing and adapter trimming with Cutadapt, quality assessment through FastQC and MultiQC, filtering with PRINSEQ, and assembly of paired reads with PANDAseq. DADA2 could generate amplicon sequence variants, while CD-HIT supported dereplication in the operational taxonomic unit, or OTU, variant of the pipeline. Swarm was available for OTU generation, VSEARCH was used to detect chimeras, and AmpliconDuo could merge samples split across sequencing runs. Taxonomic assignments were then made with BLASTn against SILVA or the broad National Center for Biotechnology Information nucleotide database. The final output combines sequence information, abundance estimates and taxonomic labels.

Natrix2 changes four major parts of that system. It adds VSEARCH as an alternative to Swarm for de novo clustering of Illumina reads, introduces mothur as a second taxonomic classifier, extends the available reference resources to include the Protist Ribosomal Reference database, known as PR2, and the fungal UNITE database, and introduces a complete processing path for Oxford Nanopore reads. These additions broaden the range of marker genes and organisms that can be studied while preserving the original pipeline’s adjustable structure. Users can select the relevant clustering and classification options through configuration files, and the project provides example sequences, workflows and instructions for running the system.

The new VSEARCH option groups Illumina sequences into OTUs through a greedy heuristic method with a centroid-based approach. Its similarity threshold is adjustable and defaults to 0.98, meaning that reads sharing 98% sequence similarity are clustered together. This provides researchers with a flexible alternative to the existing Swarm procedure. Natrix2 also incorporates mumu as a post-clustering step intended to improve the reliability of OTU generation for both Illumina and Nanopore data. The algorithm evaluates sequence similarity together with patterns of co-occurrence among reads, allowing likely erroneous clusters to be removed and helping the final OTUs represent biological diversity more accurately.

Taxonomic classification gains a new route through the classify.seqs function in the mothur package. Rather than first constructing alignments between sample sequences and database entries, mothur uses a k-mer-based procedure in which short subsequences contribute to the probability that a query belongs to a particular taxon. The sequence receiving the highest probability is assigned to that group. This alternative can be useful as reference libraries and the volume of sequencing data continue to grow. Natrix2 allows mothur classification for both Illumina and Nanopore reads, while retaining BLAST as an option.

The expanded database suite is especially important because a classifier can only assign the organisms represented in its reference collection. SILVA contains ribosomal sequences relevant to prokaryotic and eukaryotic research, and NCBI provides broad coverage across many species. PR2 extends support for 18S rRNA metabarcoding and includes protists as well as fungal, animal and plant sequences curated by domain experts. UNITE focuses on the eukaryotic nuclear ribosomal internal transcribed spacer, or ITS, region and contains more than one million fungal reference sequences. Their inclusion makes the workflow more suitable for fungal metabarcoding and for studies targeting biological groups beyond bacteria and archaea, including projects that previously would have required a separate analysis route.

For Nanopore data, Natrix2 begins with FASTQ files and uses Pychopper, an Oxford Nanopore Technologies tool, to identify and remove sequencing adapters and primer sequences. Pychopper combines global and local sequence alignments to trim reads and orient them into the forward direction. The tool also produces a file of fused reads that can undergo a secondary trimming and orientation procedure designed to rescue additional sequence information. These steps are important because reliable downstream clustering and consensus generation depend on reads having a consistent orientation and unnecessary terminal sequence having been removed.

The cleaned Nanopore reads are converted to FASTA format and clustered with CD-HIT using a similarity threshold. Representative sequences are then mapped against the original FASTA reads with Minimap2, allowing the workflow to connect each cluster to the reads that contributed to it. Racon and Medaka generate higher-quality consensus sequences from that information. Racon uses distance- and quality-based alignment to refine the consensus, whereas Medaka applies a neural-network method to correct errors. Minimap2 maps the finished consensus sequences back to the original reads to identify corresponding read numbers, and VSEARCH’s uchime3_denovo algorithm removes suspected chimeric sequences. The corrected reads can then be classified with either BLAST or mothur and incorporated into the broader Natrix analysis.

The update reflects a broader change in environmental sequencing: sequencing platforms and analytical tools are advancing faster than many fixed pipelines can accommodate. Nanopore reads may span more than 800,000 base pairs, compared with an Illumina maximum of about 300 base pairs, but their reported error rates are substantially higher. Long reads can reveal sequence features hidden by fragmentation, while dedicated correction and consensus tools improve their reliability. Natrix2 does not remove the need for careful experimental design or quality assessment; rather, it integrates the computational steps required to turn technology-specific reads into comparable sequence variants and OTUs. By making these tools available in one configurable, open-source workflow under an MIT licence, the developers aim to lower the technical barrier to studying microbial, fungal, protist, plant and animal communities with either short- or long-read amplicon data.

Subject of Research: An open-source workflow for processing and classifying Illumina and Oxford Nanopore amplicon sequences.

Article Title: Natrix2 – Improved amplicon workflow with novel Oxford Nanopore Technologies support and enhancements in clustering, classification and taxonomic databases

Article References: Deep, A., Bludau, D., Welzel, M., Clemens, S., Heider, D., Boenigk, J., & Beisser, D. (2023). Natrix2 – Improved amplicon workflow with novel Oxford Nanopore Technologies support and enhancements in clustering, classification and taxonomic databases. Metabarcoding and Metagenomics, 7, Article e109389. https://doi.org/10.3897/mbmg.7.109389

Image Credits: AI Generated

DOI: 10.3897/mbmg.7.109389

Keywords: Natrix2, metabarcoding, Oxford Nanopore, Illumina sequencing, amplicon sequencing, taxonomic classification, OTU clustering, VSEARCH, mothur, environmental DNA, Improved, amplicon