16s workflow

Taxonomic classification of single reads from amplicon-targeted sequencing.

Introduction

This workflow can be used for the following:

• Taxonomic classification of 16S rDNA and 18S rDNA amplicons using default or custom databases. Default databases:
  • NCBI targeted loci: 16S rDNA, 18S rDNA, ITS (ncbi_16s_18s, ncbi_16s_18s_28s_ITS; see here for details).
• Generate taxonomic profiles of one or more samples.

Additional features:

• Two different approaches are available: minimap2 (using alignment, default option) or kraken2 (k-mer based).
• Results include:
  • An abundance table with counts per taxa in all the samples.
  • Interactive sankey and sunburst plots to explore the different identified lineages.
  • A bar plot comparing the abundances of the most abundant taxa in all the samples.

Compute requirements

Recommended requirements:

• CPUs = 12
• Memory = 32GB

Minimum requirements:

• CPUs = 6
• Memory = 16GB

Approximate run time: ~40min for 1 million reads in total (24 barcodes) using Minimap2 and the ncbi_16s_18s database.

ARM processor support: True

Install and run

These are instructions to install and run the workflow on command line. You can also access the workflow via the EPI2ME Desktop application.

The workflow uses Nextflow to manage compute and software resources, therefore Nextflow will need to be installed before attempting to run the workflow.

The workflow can currently be run using either Docker or Singularity to provide isolation of the required software. Both methods are automated out-of-the-box provided either Docker or Singularity is installed. This is controlled by the -profile parameter as exemplified below.

It is not required to clone or download the git repository in order to run the workflow. More information on running EPI2ME workflows can be found on our website.

The following command can be used to obtain the workflow. This will pull the repository in to the assets folder of Nextflow and provide a list of all parameters available for the workflow as well as an example command:

nextflow run epi2me-labs/wf-16s --help

To update a workflow to the latest version on the command line use the following command:

nextflow pull epi2me-labs/wf-16s

A demo dataset is provided for testing of the workflow. It can be downloaded and unpacked using the following commands:

wget https://ont-exd-int-s3-euwst1-epi2me-labs.s3.amazonaws.com/wf-16s/wf-16s-demo.tar.gz
tar -xzvf wf-16s-demo.tar.gz

The workflow can then be run with the downloaded demo data using:

nextflow run epi2me-labs/wf-16s \
    --fastq 'wf-16s-demo/test_data' \
    --minimap2_by_reference \
    -profile standard

For further information about running a workflow on the command line see https://labs.epi2me.io/wfquickstart/

Related protocols

This workflow is designed to take input sequences that have been produced from Oxford Nanopore Technologies devices using this protocol:

https://community.nanoporetech.com/docs/prepare/library_prep_protocols/rapid-sequencing-DNA-16s-barcoding-kit-v14-sqk-16114-24

Find related protocols in the Nanopore community.

Input example

This workflow accepts either FASTQ or BAM files as input.

The FASTQ or BAM input parameters for this workflow accept one of three cases: (i) the path to a single FASTQ or BAM file; (ii) the path to a top-level directory containing FASTQ or BAM files; (iii) the path to a directory containing one level of sub-directories which in turn contain FASTQ or BAM files. In the first and second cases (i and ii), a sample name can be supplied with --sample. In the last case (iii), the data is assumed to be multiplexed with the names of the sub-directories as barcodes. In this case, a sample sheet can be provided with --sample_sheet.

(i)                     (ii)                 (iii)    
input_reads.fastq   ─── input_directory  ─── input_directory
                        ├── reads0.fastq     ├── barcode01
                        └── reads1.fastq     │   ├── reads0.fastq
                                             │   └── reads1.fastq
                                             ├── barcode02
                                             │   ├── reads0.fastq
                                             │   ├── reads1.fastq
                                             │   └── reads2.fastq
                                             └── barcode03
                                              └── reads0.fastq

Input parameters

Input Options

Real Time Analysis Options

Sample Options

Reference Options

Kraken2 Options

Minimap2 Options

Report Options

Output Options

Advanced Options

Outputs

Output files may be aggregated including information for all samples or provided per sample. Per-sample files will be prefixed with respective aliases and represented below as {{ alias }}.

Pipeline overview

1. Concatenate input files and generate per read stats

fastcat is used to concatenate input FASTQ files prior to downstream processing of the workflow. It will also output per-read stats including read lengths and average qualities.

You may want to choose which reads are analysed by filtering them using these flags max_len, min_len, min_read_qual, (see the Inputs section for details).

2. Remove host sequences (optional)

We have included an optional filtering step to remove any host sequences that map (using Minimap2) against a provided host reference (e.g. human), which can be a FASTA file or a MMI index. To use this option provide the path to your host reference with the exclude_host parameter. The mapped reads are output in a BAM file and excluded from further analysis.

nextflow run epi2me-labs/wf-16s --fastq test_data/case04/reads.fastq.gz --exclude_host test_data/case04/host.fasta.gz

3. Classify reads taxonomically

There are two different approaches to taxonomic classification:

3.1 Using Minimap2

Minimap2 provides better resolution, but, depending on the reference database used, can take significantly more time. Also, running the workflow with minimap2 does not support real-time analysis. This is the option by default.

nextflow run epi2me-labs/wf-16s --fastq test_data/case01 --classifier minimap2

The creation of alignment statistics plots can be enabled with the minimap2_by_reference flag. Using this option produces a table and scatter plot in the report showing sequencing depth and coverage of each reference. The report also contains a heatmap indicating the sequencing depth over relative genomic coordinates for the references with the highest coverage (references with a mean coverage of less than 1% of the one with the largest value are omitted).

In addition, the user can output BAM files in a folder called bams by using the option keep_bam. If the user provides a custom database and uses the igv option, the workflow will also output the references with reads mappings, as well as an IGV configuration file. This configuration file allows the user to view the alignments in the EPI2ME Desktop Application in the Viewer tab. Note that the number of references can be reduced using the abundance_threshold option, which will select those references with a number of reads aligned higher than this value. Please, consider that the view of the alignment is highly dependent on the reference selected.

3.2 Using Kraken2

Kraken2 provides the fastest method for the taxonomic classification of the reads. Then, Bracken is used to provide an estimate of the species (or the selected taxonomic rank) abundance in the sample.

4. Prepare output

The main output of the wf-16s pipeline is the wf-16s-report.html which can be found in the output directory. It contains a summary of read statistics, the taxonomic composition of the sample and some diversity metrics. The results shown in the report can also be customised with several options. For example, you can use abundance_threshold to remove all taxa less prevalent than the threshold from the abundance table. When setting this parameter to a natural number, taxa with fewer absolute counts are removed. You can also pass a decimal between 0.0-1.0 to drop taxa of lower relative abundance. Furthermore, n_taxa_barplot controls the number of taxa displayed in the bar plot and groups the rest under the category ‘Other’.

The workflow output also contains Kraken and bracken reports for each sample. Additionally, the ‘species-abundance.tsv’ is a table with the counts of the different taxa per sample. You can use the flag include_kraken2_assignments to include a per sample TSV file that indicates how each input sequence was classified as well as the taxon that has been assigned to each read. This TSV file will only be output on completion of the workflow and therefore not at all if using the real time option whilst running indefinitely. This option is available in the Kraken2 pipeline.

5.1 Diversity indices

Species diversity refers to the taxonomic composition in a specific microbial community. There are some useful concepts to take into account:

• Richness: number of unique taxonomic groups present in the community,
• Taxonomic group abundance: number of individuals of a particular taxonomic group present in the community,
• Evenness: refers to the equitability of the different taxonomic groups in terms of their abundances. Two different communities can host the same number of different taxonomic groups (i.e. they have the same richness), but they can have different evenness. For instance, if there is one taxon whose abundance is much larger in one community compared to the other.

There are three types of biodiversity measures described over a special scale ^{1, 2}: alpha-, beta-, and gamma-diversity.

• Alpha-diversity refers to the richness that occurs within a community given area within a region.
• Beta-diversity defined as variation in the identities of species among sites, provides a direct link between biodiversity at local scales (alpha diversity) and the broader regional species pool (gamma diversity).
• Gamma-diversity is the total observed richness within an entire region.

To provide a quick overview of the alpha-diversity of the microbial community, we provide some of the most common diversity metrics calculated for a specific taxonomic rank ³, which can be chosen by the user with the taxonomic_rank parameter (‘D’=Domain,‘P’=Phylum, ‘C’=Class, ‘O’=Order, ‘F’=Family, ‘G’=Genus, ‘S’=Species). By default, the rank is ‘G’ (genus-level). Some of the included alpha diversity metrics are:

• Shannon Diversity Index (H): Shannon entropy approaches zero if a community is almost entirely made up of a single taxon.

H = -\sum_{i=1}^{S}p_i*ln(p_i)

• Simpson’s Diversity Index (D): The range is from 0 (low diversity) to 1 (high diversity).

D = \sum_{i=1}^{S}p_i^2

• Pielou Index (J): The values range from 0 (presence of a dominant species) and 1 (maximum evennes).

J = H/ln(S)

• Berger-Parker dominance index (BP): expresses the proportional importance of the most abundant type, i.e., the ratio of number of individuals of most abundant species to the total number of individuals of all the species in the sample.

BP = n_i/N

where n_i refers to the counts of the most abundant taxon and N is the total of counts.

• Fisher’s alpha: Fisher (see Fisher, 1943⁴) noticed that only a few species tend to be abundant while most are represented by only a few individuals (‘rare biosphere’). These differences in species abundance can be incorporated into species diversity measurements such as the Fisher’s alpha. This index is based upon the logarithmic distribution of number of individuals of different species.

S = \alpha * ln(1 + N/\alpha)

where S is the total number of taxa, N is the total number of individuals in the sample. The value of Fisher’s $\alpha$ is calculated by iteration.

These indices are calculated by default using the original abundance table (see McMurdie and Holmes⁵, 2014 and Willis⁶, 2019). If you want to calculate them from a rarefied abundance table (i.e. all the samples have been subsampled to contain the same number of counts per sample, which is the 95% of the minimum number of total counts), you can download the rarefied table from the report.

The report also includes the rarefaction curve per sample which displays the mean of species richness for a subsample of reads (sample size). Generally, this curve initially grows rapidly, as most abundant species are sequenced and they add new taxa in the community, then slightly flattens due to the fact that ‘rare’ species are more difficult of being sampled, and because of that is more difficult to report an increase in the number of observed species.

Note: Within each rank, each named taxon is a unique unit. The counts are the number of reads assigned to that taxon. All Unknown sequences are considered as a unique taxon

6. Running wf-metagenomics in real time

This feature is only available when using Kraken2 as the classifier. It is somewhat experimental and may not work as expected in all environments.

The Kraken2 mode of the workflow can be used in real-time, allowing the workflow to run parallel with an ongoing sequencing run as read data is being produced by the Oxford Nanopore Technologies sequencing instrument. In this case, Kraken2 is used with the Kraken2-server and the user can visualise the classification of reads and species abundances in a real-time updating report. In real-time mode, the workflow processes new input files as they become available in batches of the specified size. Thus, this option cannot be used with a single fastq as input.

When using the workflow in real-time, the workflow will run indefinitely until a user interrupts the program (e.g with ctrl+c when on the command line). The workflow can be configured to complete automatically after a set number of reads have been analysed using the read_limit variable. Once this threshold has been reached, the program will emit a STOP.fastq.gz file into the fastq directory, which will instruct the workflow to complete. The “STOP.fastq.gz” file is then deleted.

nextflow run epi2me-labs/wf-16s --fastq test_data/case01 --real_time --batch_size 1000 --read_limit 4000

If running the Kraken2 pipeline real_time in a cluster, there are two options to enable the workflow to be able to communicate with the Kraken-server:

1. Run a Kraken-server separately outside of the workflow.
2. Submit the workflow job to run on a single node (i.e. running as if on a single local machine).

The real-time subworkflow uses a server process to handle Kraken2 classification requests. This allows the workflow to persist the sequence database in memory throughout the duration of processing. There are some parameters that may be worth considering to improve the performance of the workflow:

• port: The option specifies the local network port on which the server and clients will communicate.
• host: Network hostname (or IP address) for communication between Kraken2 server and clients. (See also external_kraken2 parameter).
• external_kraken2: Whether a pre-existing Kraken2 server should be used, rather than creating one as part of the workflow. By default the workflow assumes that it is running on a single host computer, and further that it should start its own Kraken2 server. It may be desirable to start a Kraken2 server outside of the workflow (for example to host a large database), in which case this option should be enabled. This option may be used in conjuction with the host option to specify that the Kraken2 server is running on a remote computer.
• server_threads: Number of CPU threads used by the Kraken2 server for classifying reads.
• kraken_clients: Number of clients that can connect at once to the Kraken-server for classifying reads. It should not be set to more than 4 fewer than the executor CPU limit.

Troubleshooting

• If the workflow fails please run it with the demo data set to ensure the workflow itself is working. This will help us determine if the issue is related to the environment, input parameters or a bug.
• See how to interpret some common nextflow exit codes here.
• When using the Minimap2 pipeline with a custom database, you must make sure that the ref2taxid and reference files are coherent, as well as the taxonomy database.
• If your device doesn’t have the resources to use large Kraken2 databases, you can enable kraken2_memory_mapping to reduce the amount of memory required.
• To enable IGV viewer with a custom reference, this must be a FASTA file and not a minimap2 MMI format index.

FAQ’s

If your question is not answered here, please report any issues or suggestions on the github issues page or start a discussion on the community.

• Which database is used per default? - By default, the workflow uses the NCBI 16S + 18S rRNA database. It will be downloaded the first time the workflow is run and re-used in subsequent runs.

• Are more databases available? - Other 16s databases (listed below) can be selected with the database_set parameter, but the workflow can also be used with a custom database if required (see here for details).

  • 16S, 18S, ITS
    • ncbi_16s_18s and ncbi_16s_18s_28s_ITS: Archaeal, bacterial and fungal 16S/18S and ITS data. There are two databases available using the data from [NCBI]https://www.ncbi.nlm.nih.gov/refseq/targetedloci/)
    • SILVA_138_1: The SILVA database (version 138) is also available. Note that SILVA uses its own set of taxids, which do not match the NCBI taxids. We provide the respective taxdump files, but if you prefer using the NCBI ones, you can create them from the SILVA files (NCBI). As the SILVA database uses genus level, the last taxonomic rank at which the analysis is carried out is genus (taxonomic_rank G).

• How can I use Kraken2 indexes? - There are different databases available here.

• How can I use custom databases? - If you want to run the workflow using your own Kraken2 database, you’ll need to provide the database and an associated taxonomy dump. For a custom Minimap2 reference database, you’ll need to provide a reference FASTA (or MMI) and an associated ref2taxid file. For a guide on how to build and use custom databases, take a look at our article on how to run wf-16s offline.

• How can I run the workflow with less memory? - When running in Kraken mode, you can set the kraken2_memory_mapping parameter if the available memory is smaller than the size of the database.

• How can I run the workflow offline? - To run wf-16s offline you can use the workflow to download the databases from the internet and prepare them for offline re-use later. If you want to use one of the databases supported out of the box by the workflow, you can run the workflow with your desired database and any input (for example, the test data). The database will be downloaded and prepared in a directory on your computer. Once the database has been prepared, it will be used automatically the next time you run the workflow without needing to be downloaded again. You can find advice on picking a suitable database in our article on selecting databases for wf-metagenomics.

• Is it possible to run the real time approach in the cloud? - No, the real time pipeline is not yet available to be run in the cloud.

Related blog posts

• How to build and use databases to run wf-metagenomics and wf-16s offline.
• Selecting the correct databases in the wf-metagenomics.

See the EPI2ME website for lots of other resources and blog posts.