nf-core/taxprofiler

You will need to create a design file with information about the samples and libraries/runs you want to running in your pipeline run. Use this parameter to specify its location. It has to be a comma-separated file with 6 columns, and a header row. See usage docs.

You will need to create a design file with information about the samples in your experiment before running the pipeline. Use this parameter to specify its location. It has to be a comma-separated file with 4 columns, and a header row. See usage docs.

Profilers will only be executed if a corresponding database are supplied.

We recommend storing this database sheet somewhere centrally and accessible by others members of your lab/institutions, as this file will likely be regularly reused.

If input databases are supplied as gzipped TAR archives, in some cases you may want to move and re-use these for future runs. Specifying this parameter will save these to --outdir results/ under a directory called untar.

Set this parameter to your e-mail address to get a summary e-mail with details of the run sent to you when the workflow exits. If set in your user config file (~/.nextflow/config) then you don't need to specify this on the command line for every run.

Falco is designed as a drop-in replacement for FastQC but written in C++ for faster computation. We particularly recommend using falco when using long reads (due to reduced memory constraints), however is also applicable for short reads.

This saves the FASTQ output from the following tools:

• fastp
• AdapterRemoval
• Porechop
• Filtlong

These reads will be a mixture of: adapter clipped, quality trimmed, pair-merged, and length filtered, depending on the parameters you set.

This flag will generate the directory results/analysis_read_reads that contains the reads from the last preprocessing (QC, host removal, run merging etc.) step of the pipeline run.

This can be useful if you wish to re-use the final cleaned-up and prepared reads - the data actually used for the actual classification/profiling steps of the pipeline - for other analyses or purposes (e.g., to reduce redundant preprocessing between different pipelines, e.g. nf-core/mag).

In most cases this will be preferred over similar parameters e.g. --save_preprocessed_reads or --save_complexityfiltered_reads, unless you wish to explore in more detail the output of each specific preprocessing step independently.

Note if you do no preprocessing of any kind, nothing will be present in this directory.

Turns on short read quality control steps (adapter clipping, complexity filtering etc.)

This subworkflow can perform:

• Adapter removal
• Read quality trimming
• Read pair merging
• Length filtering
• Complexity filtering

Either with fastp or AdapterRemoval.

Removing adapters (if present) is recommend to reduce false-postive hits that may occur from 'dirty' or 'contaminated' reference genomes in a profiling database that contain accidentially incorporated adapter sequences. Note that some, but not all, tools support paired-end alignment (utilising information about the insert covered by the pairs). However read pair merging in some cases can be recommend to increase read length (such as in aDNA). Length filtering, and/or complexity can speed up alignment by reducing the number of short unspecific reads that need to be aligned.

Skip the removal of sequencing adapters.

This often can be useful to speed up run-time of the pipeline when analysing data downloaded from public databases such as the ENA or SRA, as adapters should already be removed (however we recommend to check FastQC results to ensure this is the case).

Specify a custom forward or R1 adapter sequence to be removed from reads.

If not set, the selected short-read QC tool's defaults will be used.

Modifies tool parameter(s):

• fastp: --adapter_sequence. fastp default: AGATCGGAAGAGCACACGTCTGAACTCCAGTCA
• AdapterRemoval: --adapter1. AdapteRemoval2 default: AGATCGGAAGAGCACACGTCTGAACTCCAGTCACNNNNNNATCTCGTATGCCGTCTTCTGCTTG

Specify a custom reverse or R2 adapter sequence to be removed from reads.

If not set, the selected short-read QC tool's defaults will be used.

Modifies tool parameter(s):

• fastp: --adapter_sequence. fastp default: AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT
• AdapterRemoval: --adapter1. AdapteRemoval2 default: AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGTAGATCTCGGTGGTCGCCGTATCATT

Allows to supply a file with a list of adapter (combinations) to remove from all files.

Overrides the --shortread_qc_adapter1/--shortread_qc_adapter2 parameters .

For AdapterRemoval this consists of a two column table with a .txt extension: first column represents forward strand, second column for reverse strand. You must supply all possible combinations, one per line, and this list is applied to all files. See AdapterRemoval documentation for more information.

For fastp this consists of a standard FASTA format with a .fasta/.fa/.fna/.fas extension. The adapter sequence in this file should be at least 6bp long, otherwise it will be skipped. fastp trims the adapters present in the FASTA file one by one.

Modifies AdapterRemoval parameter: --adapter-list
Modifies fastp parameter: --adapter_fasta

Turn on the merging of read-pairs of paired-end short read sequencing data.

Modifies tool parameter(s):

• AdapterRemoval: --collapse
• fastp: -m --merged_out

Turns on the inclusion of unmerged reads in resulting FASTQ file from merging paired-end sequencing data when using fastp and/or AdapterRemoval. For fastp this means the unmerged read pairs are directly included in the output FASTQ file. For AdapterRemoval, additional output files containing unmerged reads are all concatenated into one file by the workflow.

Excluding unmerged reads can be useful in cases where you prefer to have very short reads (e.g. aDNA), thus excluding longer-reads or possibly faulty reads where one of the pair was discarded.

Adds fastp option: --include_unmerged

Specifying a mimum read length filtering can speed up profiling by reducing the number of short unspecific reads that need to be match/aligned to the database.

Modifies tool parameter(s):

• removed from reads --length_required
• AdapterRemoval: --minlength

This enables the deduplication of processed reads during fastp adapter removal and/or merging. It removes identical reads that are likely artefacts from laboratory protocols (e.g. amplification), and provide no additional sequence information to the library.

Removing duplicates can increase runtime and increase accuracy of abundance calculations.

Modifies tool parameter(s):
fastp: --dedup

Turns on sequencing complexity filtering. Complexity filtering can be useful to increase run-time by removing unspecific read sequences that do not provide any informative taxon ID.

Specify the minimum 'entropy' value for complexity filtering for BBDuk or PRINSEQ++.

Note that this value will only be used for PRINSEQ++ if --shortread_complexityfilter_prinseqplusplus_mode is set to entropy.

Entropy here corresponds to the amount of sequence variation exists within the read. Higher values correspond to more variety, and thus will likely reslut in more specific matching to a taxon's reference genome. The trade off here is fewer reads (or abundance information) available for having a confident identification.

Modifies tool parameter(s):

• BBDuk: entropy=
• PRINSEQ++: -lc_entropy

Specify the window size to calculate the level entropy within for BBDuk.

Modifies tool parameter(s):

• BBDuk: entropywindow=

Turn on masking of low-complexity reads (i.e., replacement with N) rather than removal.

Modifies tool parameter(s)

• BBDuk: entropymask=

Specify the minimum sequence complexity value for fastp. This value corresponds to the percentage of bases that is different from it's adjacent bases.

Modifies tool parameter(s):

• removed from reads --complexity_threshold

Specify the minimum dust score below which low-complexity reads will be removed. A DUST score is based on how often different tri-nucleotides occur along a read.

Modifies tool parameter(s):

• PRINSEQ++: --lc_dust

Specify whether to save the final complexity filtered reads in your results directory (--outdir).

Turns on long read quality control steps (adapter clipping, length and/or quality filtering.)

Removing adapters (if present) is recommend to reduce false-postive hits that may occur from 'dirty' or 'contaminated' reference genomes in a profiling database that contain accidentially incorporated adapter sequences.

Length filtering, and quality filtering can speed up alignment by reducing the number of unspecific reads that need to be aligned.

Skip removal of adapters by Porechop. This can be useful in some cases to speed up run time - particularly when you are running data downloading from public databases such as the ENA/SRA that should already have adapters removed. We recommend that you check your FastQC results this is indeed the case.

Skip removal of quality filtering with Filtlong. This will skip length, percent reads, and target bases filtering (see other --longread_qc_qualityfilter_* parameters).

Specify the minimum of length of reads to be kept for downstream analysis.

Modifies tool parameter(s):

• Filtlong: --min_length

Throw out the remaining percentage of reads outside the value. This is measured by bp, not by read count. So this option throws out the worst e.g. 10% of read bases if the parameter is set to 90. Modified from Filtlong documentation

Modifies tool parameter(s):

• Filtlong: --keep_percent

Removes the worst reads until only the specified value of bases remain, useful for very large read sets. If the input read set is less than the specified value, this setting will have no effect. Modified from Filtlong documentation

Modifies tool parameter(s):

• Filtlong: --keep_percent

Turns on the ability to remove short-reads from the that derived from a known organism, using Bowtie2 and samtools

This subworkflow is useful to remove reads that may come from a host, or a known contamination like the human reference genome. Human DNA contamination of (microbial) reference genomes is well known, so removal of these prior profiling both reduces the risks of false positives, and in some cases a faster runtime (as less reads need to be profiled).

Alternatively, you can include the reference genome within your profiling databases and can turn off this subworkflow, with the trade off of a larger taxonomic profiling database.

Turns on the ability to remove long-reads from the that derived from a known organism, using minimap2 and samtools

This subworkflow is useful to remove reads that may come from a host, or a known contamination like the human reference genome. Human DNA contamination of (microbial) reference genomes is well known, so removal of these prior profiling both reduces the risks of false positives, and in some cases a faster runtime (as less reads need to be profiled).

Alternatively, you can include the reference genome within your profiling databases and can turn off this subworkflow, with the trade off of a larger taxonomic profiling database.

Specify a path to the FASTA file (optionally gzipped) of the reference genome of the organism to be removed.

If you have two or more host organisms or contaminants you wish to remove, you can concatenate the FASTAs of the different taxa into a single one to provide to the pipeline.

Specify the path to a directory containing pre-made Bowtie2 reference index files (i.e. the directory containing .bt1, .bt2 files etc.). These should sit in the same directory alongside the the reference file specified in --hostremoval_reference.

Specifying premade indices can speed up runtime of the host-removal step, however if not supplied the pipeline will generate the indices for you.

Specify path to a pre-made Minimap2 index file (.mmi) of the host removal reference file given to --hostremoval_reference.

Specifying a premade index file can speed up runtime of the host-removal step, however if not supplied the pipeline will generate the indices for you.

Save the output files of the in-built indexing of the host genome.

This is recommend to be turned on if you plan to use the same reference genome multiple times, as supplying the directory or file to --shortread_hostremoval_index or --longread_hostremoval_index respectively can speed up runtime of future runs. Once generated, we recommend you place this file outside of your run results directory in a central 'cache' directory you and others using your machine can access and supply to the pipeline.

Save the reads mapped to the reference genome and off-target reads in BAM format as output by the respective hostremoval alignment tool.

This can be useful if you wish to perform other analyses on the host organism (such as host-microbe interaction), however, you should consider whether the default mapping parameters of Bowtie2 (short-read) or minimap2 (long-read) are optimised to your context.

Save only the reads NOT mapped to the reference genome in FASTQ format (as exported from samtools view and fastq).

This can be useful if you wish to perform other analyses on the off-target reads from the host mapping, such as manual profiling or de novo assembly.

Turns on the concatenation of sequencing runs or libraries with the same sample name.

This can be useful to ensure you get a single profile per sample, rather than one profile per run or library. Note that in some cases comparing profiles of independent libraries may be useful, so this parameter may not always be suitable.

Save the run- and library-concatenated reads of a given sample in FASTQ format.

⚠️ Only samples that went through the run-merging step of the pipeline will be stored in the resulting directory.

If you wish to save the files that go to the classification/profiling steps for samples that did not go through run merging, you must supply the appropriate upstream --save_<preprocessing_step> flag.

Save mapped (SAM, FASTQ) and unmapped (FASTQ) reads from alignment step of centrifuge in your output results directory.

Modifies tool parameter(s):

• centrifuge: --un-gz, --al-gz, --un-conc-gz, --al-conc-gz, --out-fmt

DIAMOND can produce output in a number of different formats, you can specify here which to produce.

Note that DIAMOND can only produce one format at a time, and depending on which you pick, some downstream steps may not be executed. For example, selecting daa or sam will mean you will not get a tabular taxonomic profile as with the other tools.

Will be overriden by --diamond_save_reads.

Modifies tool parameter(s):

• diamond blastx: --outfmt

Save aligned reads in SAM format from alignment step of DIAMOND in your output results directory.

Note this explicitly overrides --diamond_output_format to produce the SAM file, and no taxon table will be generated.

Modifies tool parameter(s):

• DIAMOND: --outfmt

Turn on the reporting by Kaiju of viruses at specific virus levels, rather than aggregating at specific taxonomic levels as specified by -- kaiju_taxon_rank (i.e., read counts will not be summarised at higher taxonomic levels).

Modifies tool parameter(s):

• kaiju2table: -e

Specify the taxonomic level(s) to be displayed in the resulting Kaiju taxon table, as generated by the kaiju2table helper tool.

This can be only be a single level (e.g. species).

Modifies tool parameter(s):

• kaiju2table: -l

Save reads that do and do not have a taxonomic classification in your output results directory in FASTQ format.

Modifies tool parameter(s):

• kraken2: --classified-out and --unclassified-out

Save a text file that contains a list of each read that had a taxonomic assignment, with information on specific taxonomic taxonomic assignment that that read recieved.

Modifies tool parameter(s):

• kraken2: --output

Turn on saving minimizer information in the kraken2 report thus increasing to an eight column layout.

Adds --report-minimizer-data to the kraken2 command.

Save reads that do and do not have a taxonomic classification in your output results directory in FASTA format. Reads from paired-end input are merged.

Modifies tool parameter(s):

• krakenuniq: --classified-out and --unclassified-out

nf-core/taxprofiler utilises a 'low memory' option for KrakenUniq that can reduce the amount of RAM the process requires using the --preloaded option.

A further extension to this option is that you can specify how large each chunk of the database should be that gets loaded into memory at any one time. You can specify the amount of RAM to chunk the database to with this parameter, and is particularly useful for people with limited computational resources.

More information about this parameter can be seen here.

Modifies KrakenUniq parameter: --preload-size

Save a text file that contains a list of each read that had a taxonomic assignment, with information on specific taxonomic taxonomic assignment that that read received.

Modifies tool parameter(s):

• krakenuniq: --output

Specify the batch size for KrakenUniq. The reference database for KrakenUniq is loaded into memory once per nextflow process and then used to classify many samples. When you have many samples, a single KrakenUniq run can be rather slow. Alternatively, we can split up KrakenUniq runs for a 'batch' of samples, allowing a balance between shared use of database for multiple samples, but also faster parallelised KrakenUniq runs. This parameter determines for how many samples at a time.

Specify which version of MALT alignment to use.

BlastN is generally recommended (nucleotide-nucleotide alignment), but particularly for very short reads (such as aDNA), whereas BlastX mode is similar to DIAMOND and will translate the nucleotide to amino acid sequences. Note each type of alignment mode requires different parameters during database construction. Refer to the MALT manual for more information.

Modifies tool parameter(s):

• malt-run: -mode

Turns on saving of MALT aligned reads in SAM format.

Note that the SAM format produce by MALT is not completely valid, and may not work with downstream tools.

Modifies tool parameter(s):

• malt-run: --alignments, -za

Turns on saving of MALT output in an additional MEGAN summary file (.megan) that can be loaded into the MEGAN metagenomic exploration tool.

Note: this file is generated not directly from MALT but rather then MEGAN utility script rma2info.

Modifies tool parameter(s):

• rma2info: -es

This parameter specifies to use the calculated relative abundance (i.e., percentage of all hits). This is normally the default behaviour in mOTUs however we use counts by default in nf-core/taxprofiler for consistency with other classification/profiling tools.

Modifies tool parameter(s):

• mOTUs: -c (removed from the default nf-core/taxprofiler command)

Turns on the saving of the read counts against each mOTU marker-gene clusters, in addition to per-taxon count/abundance reporting (in a separate file)

Modifies tool parameter(s):

• mOTUs: -M

By default mOTUs will report species names rather than NCBI Taxon IDs. In nf-core/taxprofiler we prefer taxon IDs due to interoperatbility and comparability with the output of other classifiers and profilers. If you prefer to have just species names, you can specify this to remove the IDs.

Modifies tool parameter(s):

• mOTUs: -p (removed from the default nf-core/taxprofiler command)

Available values:
0 (for pathogen detection)
1 (higherrecall)
2 (high recall)
3 (default)
4 (high precision)
5 (higher precision).
For more information about the different profiling modes, please see the kmcp documentation

Modifies tool parameter(s):

• kmcp profile: --mode

During the searching step, KMCP searches across the database with all k-mers and returns reference genome chunks sharing enough k-mers with the query. The output file is a tab-delimited file in gzip format with 15 columns and is used to generate the taxonomic profiling. More information about the columns can be found here. https://bioinf.shenwei.me/kmcp/usage/#search.

Saves .lca, .all, and .unc text files that contains a list of each read that had a taxonomic assignment, with information on specific taxonomic assignment that the read received.

Modifies tool parameter(s):

• ganon classify: --output-all --output-lca --output-unclassified

Specify the type of taxonomic report to produce from ganon report. This mainly refers to which form of 'value' to print: raw read counts, abundance estimates, genome-size normalised etc.

See the ganon documentation for more information of each option.

Modifies tool parameter(s):

• ganon report: --report-type

Specify the taxonomic rank level to report each taxonomic hit as. all will specify all ranks, however you can customise this from superkingdom through to species to as specific as assembly. ganon has a default preset, however you can customise the specific ranks in a comma separated list, e.g. --ganon_report_rank [phylum,genus,species].

See the ganon documentation for more information of each option.

Modifies tool parameter(s):

• ganon report: --ranks

Specify the top percentile or relative abundance
under which all hits underneath the threshold are not reported. This can be useful to remove long tails of few-reads and thus unconfident hits.

Modifies tool parameter(s)

• ganon report: --top-percentile

Specify the minmum number of reads or percentage of counts a hit must have against a taxon to be retained. To specify a minimum percentage, specify between 0 and 1 (e.g. 0.1 for 10%), and more than 1 to specify a hard count cut off (e.g. 100 for minimum of 100 reads).

Modifies tool parameter(s):

• ganon report: --min-count

Specify the maximum number of reads or percentage of counts a hit must have against a taxon to be retained. To specify a maximum percentage, specify between 0 and 1 (e.g. 0.9 for 90%), and more than 1 to specify a hard count cut off (e.g. 10000 for maximum of 10,000 reads).

Modifies tool parameter(s):

• ganon report: --max-count

Turns on standardisation of output OTU tables across all tools.

This happens in two forms, firstly - if available - by a given classifiers/profilers 'native' profile merger and standardisation (for Bracken, Kaiju, Kraken, Centrifuge, MetaPhlAn3, mOTUs), and secondly for all classifier/profilers in the pipeline using taxpasta.

In the latter case, taxpasta generates a standardised output as follows:

|TAXON | SAMPLE_A | SAMPLE_B |
|-------------|----------------|-----------------|
| taxon_a | 32 | 123 |
| taxon_b | 1 | 5 |

whereas all other 'native' tools have varying format outputs. See pipeline output documentation for more information.

Turn on the saving of the taxonomic output in BIOM format (.biom) in the results directory of your pipeline run, instead of the default TSV format.\n\nNote this file is from the output of the motus merge command.\n\n> Modifies tool parameter(s):\n> - -B -o

Turn on the generation of Krona interactive pie-chart HTMLs for a selection of profilers.

The tools currently supported are:

• centrifuge
• kraken2
• kaiju
• MALT

Specify a path to a Krona taxonomy database directory (i.e. a directory containing a krona generated .tab file).

This is only required for generating Krona plots of MALT output.

Note this taxonomy database must be downloaded and generated with the updateTaxonomy.sh script from the krona-tools package.

This arguments provides the path to the directory containing taxdump files. At least nodes.dmp and names.dmp are required. A merged.dmp file is optional.

Modifies tool parameter(s):
-taxpasta: --taxpasta_taxonomy_dir

The standard output format of taxpasta is a two-column table including the read counts and the integer taxonomic ID. The taxon name can be added as additional information to the output table in addition to the taxon ID.

Modifies tool parameter(s):

• taxpasta: --add-name

The standard output format of taxpasta is a two-column table including the read counts and the integer taxonomic ID. The taxon rank of the given taxonomic entry can be added as additional information to the output table.

Modifies tool parameter(s):

• taxpasta: --add-rank

The standard output format of taxpasta is a two-column table including the read counts and the integer taxonomic ID. The taxon's entire taxonomic lineage with the taxon names separated by semi-colons can be added as additional information to the output table.

Modifies tool parameter(s):

• taxpasta: --add-lineage

The standard output format of taxpasta is a two-column table including the read counts and the integer taxonomic ID. The taxon's entire taxonomic lineage with the taxon identifiers separated by semi-colons can be added as additional information to the output table.

Modifies tool parameter(s):

• taxpasta: --add-id-lineage

The standard output format of taxpasta is a two-column table including the read counts and the integer taxonomic ID. The taxonomic ranks categories of the taxon's entire lineage separated by semi-colons can be added as additional information to the output table. This complements --taxpasta_add_lineage by telling you which taxonomic rank level each entry in the lineage refers to.

Modifies tool parameter(s):

• taxpasta: --add-rank-lineage

Ignore any metagenomic profiles with errors when running taxpasta merge. At least two profiles without errors are needed to merge.

Modifies tool parameter(s):

• taxpasta: --ignore-errors