1
0
Fork 0
mirror of https://github.com/MillironX/taxprofiler.git synced 2024-11-22 09:09:55 +00:00

refactor: simplify DB validation

This commit is contained in:
Moritz E. Beber 2023-03-07 12:30:02 +01:00
parent 53db964bcf
commit c2ad421051

View file

@ -28,9 +28,9 @@ workflow DB_CHECK {
// Normal checks for within-row validity, so can be moved to separate functions // Normal checks for within-row validity, so can be moved to separate functions
parsed_samplesheet = Channel.fromPath(dbsheet) parsed_samplesheet = Channel.fromPath(dbsheet)
.splitCsv ( header:true, sep:',' ) .splitCsv ( header:true, sep:',' )
.map { .map { row ->
validate_db_rows(it) validate_db_rows(row)
create_db_channels(it) return [ row.subMap(['tool', 'db_name', 'db_params']), file(row.db_path) ]
} }
ch_dbs_for_untar = parsed_samplesheet ch_dbs_for_untar = parsed_samplesheet
@ -54,41 +54,27 @@ workflow DB_CHECK {
versions = ch_versions // channel: [ versions.yml ] versions = ch_versions // channel: [ versions.yml ]
} }
def validate_db_rows(LinkedHashMap row){ def validate_db_rows(LinkedHashMap row) {
// check minimum number of columns // 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) exit 1, "[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 // all columns there
def expected_headers = ['tool', 'db_name', 'db_params', 'db_path'] 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_keys.join(", ")}" 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_keys.join(", ")}"
// valid tools specified // valid tools specified
def expected_tools = [ "bracken", "centrifuge", "diamond", "kaiju", "kraken2", "krakenuniq", "malt", "metaphlan3", "motus" ] 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) ) exit 1, "[nf-core/taxprofiler] ERROR: Invalid tool name. Please see documentation for all supported profilers. Error in: ${row}"
// detect quotes in params // 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("'") ) 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}"
// check if any form of bracken params, that it must have `;` // 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(";") ) exit 1, "[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}"
} }
def create_db_channels(LinkedHashMap row) {
def meta = [:]
meta.tool = row.tool
meta.db_name = row.db_name
meta.db_params = row.db_params
def array = []
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}"
}
array = [ meta, file(row.db_path) ]
return array
}