def qa(self, options):
"""QA command"""
self.logger.info('[CheckM - qa] Tabulating genome statistics.')
checkDirExists(options.analyze_dir)
if options.exclude_markers:
checkFileExists(options.exclude_markers)
# calculate AAI between marks with multiple hits in a single bin
aai = AminoAcidIdentity()
aai.run(options.aai_strain, options.analyze_dir,
options.alignment_file)
# get HMM file for each bin
markerSetParser = MarkerSetParser(options.threads)
hmmModelInfoFile = os.path.join(options.analyze_dir, 'storage',
DefaultValues.CHECKM_HMM_MODEL_INFO)
binIdToModels = markerSetParser.loadBinModels(hmmModelInfoFile)
binIdToBinMarkerSets = markerSetParser.getMarkerSets(
options.analyze_dir, getBinIdsFromOutDir(options.analyze_dir),
options.marker_file, options.exclude_markers)
# get results for each bin
RP = ResultsParser(binIdToModels)
RP.analyseResults(
options.analyze_dir,
DefaultValues.BIN_STATS_OUT,
DefaultValues.HMMER_TABLE_OUT,
bIgnoreThresholds=options.bIgnoreThresholds,
evalueThreshold=options.e_value,
lengthThreshold=options.length,
bSkipPseudoGeneCorrection=options.bSkipPseudoGeneCorrection,
bSkipAdjCorrection=options.bSkipAdjCorrection)
RP.printSummary(options.out_format,
aai,
binIdToBinMarkerSets,
options.bIndividualMarkers,
options.coverage_file,
options.bTabTable,
options.file,
anaFolder=options.analyze_dir)
RP.cacheResults(options.analyze_dir, binIdToBinMarkerSets,
options.bIndividualMarkers)
if options.file != '':
self.logger.info('QA information written to: ' + options.file)
self.timeKeeper.printTimeStamp()
def treeQA(self, options):
"""QA command"""
self.logger.info(
'[CheckM - tree_qa] Assessing phylogenetic markers found in each bin.'
)
checkDirExists(options.tree_dir)
# set HMM file for each bin
markerSetParser = MarkerSetParser()
hmmModelInfoFile = os.path.join(options.tree_dir, 'storage',
DefaultValues.PHYLO_HMM_MODEL_INFO)
binIdToModels = markerSetParser.loadBinModels(hmmModelInfoFile)
# calculate marker gene statistics
RP = ResultsParser(binIdToModels)
binStats = RP.analyseResults(options.tree_dir,
DefaultValues.BIN_STATS_PHYLO_OUT,
DefaultValues.HMMER_TABLE_PHYLO_OUT)
# determine taxonomy of each bin
treeParser = TreeParser()
treeParser.printSummary(options.out_format, options.tree_dir, RP,
options.bTabTable, options.file, binStats)
if options.file != '':
self.logger.info('QA information written to: ' + options.file)
self.timeKeeper.printTimeStamp()
def lineageSet(self, options, db=None):
"""Lineage set command"""
self.logger.info(
'[CheckM - lineage_set] Inferring lineage-specific marker sets.')
checkDirExists(options.tree_dir)
# set HMM file for each bin
markerSetParser = MarkerSetParser()
hmmModelInfoFile = os.path.join(options.tree_dir, 'storage',
DefaultValues.PHYLO_HMM_MODEL_INFO)
binIdToModels = markerSetParser.loadBinModels(hmmModelInfoFile)
# calculate marker gene statistics
resultsParser = ResultsParser(binIdToModels)
resultsParser.analyseResults(options.tree_dir,
DefaultValues.BIN_STATS_PHYLO_OUT,
DefaultValues.HMMER_TABLE_PHYLO_OUT)
# These options are incompatible with how the lineage-specific marker set is selected, so
# the default values are currently hard-coded
options.num_genomes_markers = 2
options.bootstrap = 0
options.bRequireTaxonomy = False
treeParser = TreeParser()
treeParser.getBinMarkerSets(
options.tree_dir, options.marker_file, options.num_genomes_markers,
options.bootstrap, options.bNoLineageSpecificRefinement,
options.bForceDomain, options.bRequireTaxonomy, resultsParser,
options.unique, options.multi)
self.logger.info('Marker set written to: ' + options.marker_file)
self.timeKeeper.printTimeStamp()
