millironx/taxprofiler
1
0
Fork 0
mirror of https://github.com/MillironX/taxprofiler.git synced 2024-11-23 09:39:55 +00:00
Code Issues Releases Packages Activity

Merge branch 'dev' into multiqc-optimisations

Browse source
This commit is contained in:
James A. Fellows Yates 2023-04-21 05:58:23 +02:00 committed by GitHub
commit 5a09319fb3
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 21 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
CHANGELOG.md
View file

@ -13,6 +13,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
- [#269](https://github.com/nf-core/taxprofiler/pull/269/files) Reduced output files in AWS full test output due to very large files
- [#270](https://github.com/nf-core/taxprofiler/pull/270/files) Fixed warning for host removal index parameter, and improved index checks (♥ to @prototaxites for reporting, fix by @jfy133)
- [#274](https://github.com/nf-core/taxprofiler/pull/274/files) Substituted the samtools/bam2fq module with samtools/fastq module (fix by @sofstam)
- [#275](https://github.com/nf-core/taxprofiler/pull/275/files) Replaced function used for error reporting to more Nextflow friendly method (fix by @jfy133)
- [#286](https://github.com/nf-core/taxprofiler/pull/286/files) Runtime optimisation of MultiQC step via improved log file processing (fix by @Midnighter & @jfy133)
### `Dependencies`

16
subworkflows/local/db_check.nf
View file

@ -22,7 +22,7 @@ workflow DB_CHECK {
.map {
tool, db_name ->
def unique_names = db_name.unique(false)
if ( unique_names.size() < db_name.size() ) exit 1, "[nf-core/taxprofiler] ERROR: Each database for a tool must have a unique name, duplicated detected. Tool: ${tool}, Database name: ${unique_names}"
if ( unique_names.size() < db_name.size() ) error("[nf-core/taxprofiler] ERROR: Each database for a tool must have a unique name, duplicated detected. Tool: ${tool}, Database name: ${unique_names}")
}
// Normal checks for within-row validity, so can be moved to separate functions
@ -55,24 +55,24 @@ workflow DB_CHECK {
def validate_db_rows(LinkedHashMap row) {
// check minimum number of columns
if (row.size() < 4) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database input sheet - malformed row (e.g. missing column). See documentation for more information. Error in: ${row}"
if (row.size() < 4) error("[nf-core/taxprofiler] ERROR: Invalid database input sheet - malformed row (e.g. missing column). See documentation for more information. Error in: ${row}")
// all columns there
def expected_headers = ['tool', 'db_name', 'db_params', 'db_path']
if ( !row.keySet().containsAll(expected_headers) ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database input sheet - malformed column names. Please check input TSV. Column names should be: ${expected_headers.join(", ")}"
if ( !row.keySet().containsAll(expected_headers) ) error("[nf-core/taxprofiler] ERROR: Invalid database input sheet - malformed column names. Please check input TSV. Column names should be: ${expected_headers.join(", ")}")
// valid tools specified
def expected_tools = [ "bracken", "centrifuge", "diamond", "kaiju", "kraken2", "krakenuniq", "malt", "metaphlan3", "motus" ]
if ( !expected_tools.contains(row.tool) ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid tool name. Please see documentation for all supported profilers. Error in: ${row}"
if ( !expected_tools.contains(row.tool) ) error("[nf-core/taxprofiler] ERROR: Invalid tool name. Please see documentation for all supported profilers. Error in: ${row}")
// detect quotes in params
if ( row.db_params.contains('"') ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database db_params entry. No quotes allowed. Error in: ${row}"
if ( row.db_params.contains("'") ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database db_params entry. No quotes allowed. Error in: ${row}"
if ( row.db_params.contains('"') ) error("[nf-core/taxprofiler] ERROR: Invalid database db_params entry. No quotes allowed. Error in: ${row}")
if ( row.db_params.contains("'") ) error("[nf-core/taxprofiler] ERROR: Invalid database db_params entry. No quotes allowed. Error in: ${row}")
// check if any form of bracken params, that it must have `;`
if ( row.tool == 'bracken' && row.db_params && !row.db_params.contains(";") ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid database db_params entry. Bracken requires a semi-colon if passing parameter. Error in: ${row}"
if ( row.tool == 'bracken' && row.db_params && !row.db_params.contains(";") ) error("[nf-core/taxprofiler] ERROR: Invalid database db_params entry. Bracken requires a semi-colon if passing parameter. Error in: ${row}")
// ensure that the database directory exists
if (!file(row.db_path, type: 'dir').exists()) exit 1, "ERROR: Please check input samplesheet -> database path could not be found!\n${row.db_path}"
if (!file(row.db_path, type: 'dir').exists()) error("ERROR: Please check input samplesheet -> database path could not be found!\n${row.db_path}")
}

8
subworkflows/local/input_check.nf
View file

@ -44,7 +44,7 @@ def create_fastq_channel(LinkedHashMap row) {
// add path(s) of the fastq file(s) to the meta map
if (!file(row.fastq_1).exists()) {
exit 1, "ERROR: Please check input samplesheet -> Read 1 FastQ file does not exist!\n${row.fastq_1}"
error("ERROR: Please check input samplesheet -> Read 1 FastQ file does not exist!\n${row.fastq_1}")
}
if (meta.single_end) {
@ -52,12 +52,12 @@ def create_fastq_channel(LinkedHashMap row) {
} else {
if (meta.instrument_platform == 'OXFORD_NANOPORE') {
if (row.fastq_2 != '') {
exit 1, "ERROR: Please check input samplesheet -> For Oxford Nanopore reads Read 2 FastQ should be empty!\n${row.fastq_2}"
error("ERROR: Please check input samplesheet -> For Oxford Nanopore reads Read 2 FastQ should be empty!\n${row.fastq_2}")
}
return [ meta, [ file(row.fastq_1) ] ]
} else {
if (!file(row.fastq_2).exists()) {
exit 1, "ERROR: Please check input samplesheet -> Read 2 FastQ file does not exist!\n${row.fastq_2}"
error("ERROR: Please check input samplesheet -> Read 2 FastQ file does not exist!\n${row.fastq_2}")
}
return [ meta, [ file(row.fastq_1), file(row.fastq_2) ] ]
}
@ -72,7 +72,7 @@ def create_fasta_channel(LinkedHashMap row) {
meta.is_fasta = true
if (!file(row.fasta).exists()) {
exit 1, "ERROR: Please check input samplesheet -> FastA file does not exist!\n${row.fasta}"
error("ERROR: Please check input samplesheet -> FastA file does not exist!\n${row.fasta}")
}
return [ meta, [ file(row.fasta) ] ]
}

16
workflows/taxprofiler.nf
View file

@ -24,19 +24,19 @@ for (param in checkPathParamList) { if (param) { file(param, checkIfExists: true
if ( params.input ) {
ch_input = file(params.input, checkIfExists: true)
} else {
exit 1, "Input samplesheet not specified"
error("Input samplesheet not specified")
}
if (params.databases) { ch_databases = file(params.databases, checkIfExists: true) } else { exit 1, 'Input database sheet not specified!' }
if (params.databases) { ch_databases = file(params.databases, checkIfExists: true) } else { error('Input database sheet not specified!') }
if (!params.shortread_qc_mergepairs && params.run_malt ) log.warn "[nf-core/taxprofiler] MALT does not accept uncollapsed paired-reads. Pairs will be profiled as separate files."
if (params.shortread_qc_includeunmerged && !params.shortread_qc_mergepairs) exit 1, "ERROR: [nf-core/taxprofiler] cannot include unmerged reads when merging is not turned on. Please specify --shortread_qc_mergepairs"
if (params.shortread_qc_includeunmerged && !params.shortread_qc_mergepairs) error("ERROR: [nf-core/taxprofiler] cannot include unmerged reads when merging is not turned on. Please specify --shortread_qc_mergepairs")
if (params.shortread_complexityfilter_tool == 'fastp' && ( params.perform_shortread_qc == false || params.shortread_qc_tool != 'fastp' )) exit 1, "ERROR: [nf-core/taxprofiler] cannot use fastp complexity filtering if preprocessing not turned on and/or tool is not fastp. Please specify --perform_shortread_qc and/or --shortread_qc_tool 'fastp'"
if (params.shortread_complexityfilter_tool == 'fastp' && ( params.perform_shortread_qc == false || params.shortread_qc_tool != 'fastp' )) error("ERROR: [nf-core/taxprofiler] cannot use fastp complexity filtering if preprocessing not turned on and/or tool is not fastp. Please specify --perform_shortread_qc and/or --shortread_qc_tool 'fastp'")
if (params.perform_shortread_hostremoval && !params.hostremoval_reference) { exit 1, "ERROR: [nf-core/taxprofiler] --shortread_hostremoval requested but no --hostremoval_reference FASTA supplied. Check input." }
if (params.perform_shortread_hostremoval && !params.hostremoval_reference && params.shortread_hostremoval_index) { exit 1, "ERROR: [nf-core/taxprofiler] --shortread_hostremoval_index provided but no --hostremoval_reference FASTA supplied. Check input." }
if (params.perform_longread_hostremoval && !params.hostremoval_reference && params.longread_hostremoval_index) { exit 1, "ERROR: [nf-core/taxprofiler] --longread_hostremoval_index provided but no --hostremoval_reference FASTA supplied. Check input." }
if (params.perform_shortread_hostremoval && !params.hostremoval_reference) { error("ERROR: [nf-core/taxprofiler] --shortread_hostremoval requested but no --hostremoval_reference FASTA supplied. Check input.") }
if (params.perform_shortread_hostremoval && !params.hostremoval_reference && params.shortread_hostremoval_index) { error("ERROR: [nf-core/taxprofiler] --shortread_hostremoval_index provided but no --hostremoval_reference FASTA supplied. Check input.") }
if (params.perform_longread_hostremoval && !params.hostremoval_reference && params.longread_hostremoval_index) { error("ERROR: [nf-core/taxprofiler] --longread_hostremoval_index provided but no --hostremoval_reference FASTA supplied. Check input.") }
if (params.hostremoval_reference ) { ch_reference = file(params.hostremoval_reference) }
if (params.shortread_hostremoval_index ) { ch_shortread_reference_index = Channel.fromPath(params.shortread_hostremoval_index).map{[[], it]} } else { ch_shortread_reference_index = [] }
@ -45,7 +45,7 @@ if (params.longread_hostremoval_index ) { ch_longread_reference_index = fi
if (params.diamond_save_reads ) log.warn "[nf-core/taxprofiler] DIAMOND only allows output of a single format. As --diamond_save_reads supplied, only aligned reads in SAM format will be produced, no taxonomic profiles will be available."
if (params.run_malt && params.run_krona && !params.krona_taxonomy_directory) log.warn "[nf-core/taxprofiler] Krona can only be run on MALT output if path to Krona taxonomy database supplied to --krona_taxonomy_directory. Krona will not be executed in this run for MALT."
if (params.run_bracken && !params.run_kraken2) exit 1, 'ERROR: [nf-core/taxprofiler] You are attempting to run Bracken without running kraken2. This is not possible! Please set --run_kraken2 as well.'
if (params.run_bracken && !params.run_kraken2) error('ERROR: [nf-core/taxprofiler] You are attempting to run Bracken without running kraken2. This is not possible! Please set --run_kraken2 as well.')
/*
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~