def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameInputGraph = self.mConfig.get( "files", "input_graph", "pairsdb_40x40.links.gz")
self.mFilenameOutputGraph = self.mConfig.get( "files", "output_graph", "adda.graph")
self.mFilenameOutputIndex = self.mConfig.get( "files", "output_index", "adda.graph.index")
cadda.setLogLevel( self.mLogLevel )
cadda.setReportStep( self.mConfig.get( "adda", "report_step", 1000 ) )
cadda.dump_parameters()
self.mFilenames = (self.mFilenameOutputGraph,
self.mFilenameOutputIndex, )
self.mAlignmentFormat = self.mConfig.get( "files", "graph_format", "pairsdb")
if self.mAlignmentFormat == "pairsdb":
self.mGraphIterator = cadda.PairsDBNeighbourIterator
elif self.mAlignmentFormat == "pairsdb-old":
self.mGraphIterator = cadda.PairsDBNeighbourIteratorOldFormat
elif self.mAlignmentFormat == "simap":
self.mGraphIterator = AddaIO.NeighbourIteratorSimap
elif self.mAlignmentFormat == "pairsdb-realign":
self.mGraphIterator = AddaIO.NeighbourRecordPairsdbRealign
else:
raise ValueError ("unknown record type %s" % self.mAlignmentFormat)
self.info("indexing graph in format %s" % (self.mAlignmentFormat, ))
def finish( self ):
"""clean up."""
self.mOutfileFamilies.close()
self.mOutfile.close()
AddaModuleBlock.finish( self )
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenameInputGraph = self.mConfig.get("files", "input_graph",
"pairsdb_40x40.links.gz")
self.mFilenameOutputGraph = self.mConfig.get("files", "output_graph",
"adda.graph")
self.mFilenameOutputIndex = self.mConfig.get("files", "output_index",
"adda.graph.index")
cadda.setLogLevel(self.mLogLevel)
cadda.setReportStep(self.mConfig.get("adda", "report_step", 1000))
cadda.dump_parameters()
self.mFilenames = (
self.mFilenameOutputGraph,
self.mFilenameOutputIndex,
)
self.mAlignmentFormat = self.mConfig.get("files", "graph_format",
"pairsdb")
if self.mAlignmentFormat == "pairsdb":
self.mGraphIterator = cadda.PairsDBNeighbourIterator
elif self.mAlignmentFormat == "pairsdb-old":
self.mGraphIterator = cadda.PairsDBNeighbourIteratorOldFormat
elif self.mAlignmentFormat == "simap":
self.mGraphIterator = AddaIO.NeighbourIteratorSimap
elif self.mAlignmentFormat == "pairsdb-realign":
self.mGraphIterator = AddaIO.NeighbourRecordPairsdbRealign
else:
raise ValueError("unknown record type %s" % self.mAlignmentFormat)
self.info("indexing graph in format %s" % (self.mAlignmentFormat, ))
def finish(self):
"""clean up."""
self.mOutfileFamilies.close()
self.mOutfile.close()
AddaModuleBlock.finish(self)
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenameClusters = self.mConfig.get("files", "output_clusters",
"adda.clusters")
self.mFilenameAlignments = self.mConfig.get("files", "output_align",
"adda.align")
self.mFilenameNids = self.mConfig.get("files", "output_nids",
"adda.nids")
self.mFilenameGraph = self.mConfig.get("files", "output_graph",
"adda.graph")
self.mFilenameIndex = self.mConfig.get("files", "output_index",
"adda.graph.index")
self.mFilenameProfiles = self.mConfig.get("files", "output_profiles",
"adda.profiles")
self.mFilenameTransfers = self.mConfig.get("files",
"output_fit_transfer",
"adda.fit.transfer")
self.mFilenameDomains = self.mConfig.get("files", "output_domains",
"adda.domains")
self.mFilenameSummary = self.mConfig.get("files", "output_summary",
"adda.summary")
self.mFilenameDomainGraph = self.mConfig.get("files",
"output_domain_graph",
"adda.domaingraph.gz")
self.mFilenameMst = self.mConfig.get("files", "output_mst", "adda.mst")
self.mFilenameResult = self.mConfig.get("files", "output_result",
"adda.result")
self.mFilenameAlignments = self.mConfig.get("files", "output_align",
"adda.align")
self.mFilenameSegments = self.mConfig.get("files", "output_segments",
"adda.segments")
self.mFilenames = (self.mFilenameSummary, )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenamesNids = self.mConfig.get( "files", "output_nids", "adda.nids" )
self.mMinAlignedResidues = self.mConfig.get("cluster", "min_aligned_residues", 30 )
self.mPatternFamily = self.mConfig.get("cluster", "pattern_family", "AD%06i" )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameGraph = self.mConfig.get("files", "output_graph", "adda.graph" )
self.mFilenameIndex = self.mConfig.get("files", "output_index", "adda.graph.idx" )
self.mFilenameNids = self.mConfig.get("files", "output_nids", "adda.nids" )
self.mFilenameStats = self.mConfig.get("files", "output_stats", "adda.stats" )
self.mFilenameStatsSequences = self.mFilenameStats + ".persequence"
self.mFilenames = (self.mFilenameStats, self.mFilenameStatsSequences )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameOutputFasta = self.mConfig.get( "files", "output_fasta", "adda" )
self.mBlastResults = self.mConfig.get( "files", "output_blast", "adda.blast.gz" )
self.mBlastDatabase = self.mConfig.get( "blast", "database", "adda" )
self.mBlastCPUs = self.mConfig.get( "blast", "num_cpus", 2 )
self.mBlastEvalue = self.mConfig.get( "blast", "evalue", 1.0 )
self.mBlastNumResults = self.mConfig.get( "blast", "num_results", 100000 )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameNids = self.mConfig.get( "files", "output_nids", "adda.nids" )
self.mFilenameInputFasta = self.mConfig.get( "files", "input_fasta" )
self.mFilenameOutputFasta = self.mConfig.get( "files", "output_fasta", "adda" )
self.mMaxSequenceLength = self.mConfig.get( "segments", "max_sequence_length", 10000 )
self.mFilenames = (self.mFilenameNids, )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameDomainGraph = self.mConfig.get( "output", "domaingraph", "adda.domaingraph.gz" )
self.mFilenameMst = self.mConfig.get( "output", "mst", "adda.mst" )
cadda.setLogLevel( self.mLogLevel )
# cadda.setReportStep( 1 )
self.mFilenames = ( self.mFilenameMst, )
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenamesNids = self.mConfig.get("files", "output_nids",
"adda.nids")
self.mMinAlignedResidues = self.mConfig.get("cluster",
"min_aligned_residues", 30)
self.mPatternFamily = self.mConfig.get("cluster", "pattern_family",
"AD%06i")
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenameDomainGraph = self.mConfig.get("files",
"output_domaingraph",
"adda.domaingraph.gz")
self.mFilenameMst = self.mConfig.get("files", "output_mst", "adda.mst")
cadda.setLogLevel(self.mLogLevel)
# cadda.setReportStep( 1 )
self.mFilenames = (self.mFilenameMst, )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameClusters = self.mConfig.get( "output", "clusters", "adda.clusters" )
self.mFilenameFamilies = self.mConfig.get( "output", "families", "adda.families" )
self.mFilenameDomains = self.mConfig.get( "output", "adda", "adda.result" )
self.mFilenamesNids = self.mConfig.get( "output", "nids", "adda.nids" )
self.mFilenames = (self.mFilenameFamilies, self.mFilenameDomains )
self.mPatternFamily = self.mConfig.get("cluster", "pattern_family", "AD%06i" )
self.mMinDomainSize = self.mConfig.get("adda", "min_domain_size", 30 )
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenameOutputFasta = self.mConfig.get("files", "output_fasta",
"adda")
self.mBlastResults = self.mConfig.get("files", "output_blast",
"adda.blast.gz")
self.mBlastDatabase = self.mConfig.get("blast", "database", "adda")
self.mBlastCPUs = self.mConfig.get("blast", "num_cpus", 2)
self.mBlastEvalue = self.mConfig.get("blast", "evalue", 1.0)
self.mBlastNumResults = self.mConfig.get("blast", "num_results",
100000)
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenameNids = self.mConfig.get("files", "output_nids",
"adda.nids")
self.mFilenameInputFasta = self.mConfig.get("files", "input_fasta")
self.mFilenameOutputFasta = self.mConfig.get("files", "output_fasta",
"adda")
self.mMaxSequenceLength = self.mConfig.get("segments",
"max_sequence_length",
10000)
self.mFilenames = (self.mFilenameNids, )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameGraph = self.mConfig.get( "output", "graph", "adda.graph" )
self.mFilenameDomains = self.mConfig.get( "output", "domains", "adda.domains" )
self.mEvalueThresholdTrustedLinks = float(self.mConfig.get( "align", "evalue_threshold_trusted_links", -12.0 ))
self.mFilenameDomainGraph = self.mConfig.get( "output", "domaingraph", "adda.domaingraph.gz" )
cadda.setFilenameGraph( self.mFilenameGraph )
cadda.setFilenameDomains( self.mFilenameDomains )
cadda.setLogLevel( self.mLogLevel )
cadda.setEvalueThresholdTrustedLinks( self.mEvalueThresholdTrustedLinks )
self.mFilenames = (self.mFilenameDomainGraph, )
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenameClusters = self.mConfig.get("files", "output_clusters",
"adda.clusters")
self.mFilenameFamilies = self.mConfig.get("files", "output_families",
"adda.families")
self.mFilenameDomains = self.mConfig.get("files", "output_adda",
"adda.result")
self.mFilenamesNids = self.mConfig.get("files", "output_nids",
"adda.nids")
self.mFilenames = (self.mFilenameFamilies, self.mFilenameDomains)
self.mPatternFamily = self.mConfig.get("cluster", "pattern_family",
"AD%06i")
self.mMinDomainSize = self.mConfig.get("adda", "min_domain_size", 30)
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameClusters = self.mConfig.get( "output", "clusters", "adda.clusters" )
self.mFilenameAlignments = self.mConfig.get("output","align", "adda.align" )
self.mFilenameNids = self.mConfig.get( "output", "nids", "adda.nids" )
self.mFilenameGraph = self.mConfig.get( "output", "graph", "adda.graph")
self.mFilenameIndex = self.mConfig.get( "output", "index", "adda.graph.index")
self.mFilenameProfiles = self.mConfig.get( "output", "profiles", "adda.profiles" )
self.mFilenameTransfers = self.mConfig.get( "output", "fit_transfer", "adda.fit.transfer" )
self.mFilenameDomains = self.mConfig.get( "output", "domains", "adda.domains" )
self.mFilenameSummary = self.mConfig.get( "output", "summary", "adda.summary" )
self.mFilenameDomainGraph = self.mConfig.get( "output", "domain_graph", "adda.domaingraph.gz" )
self.mFilenameMst = self.mConfig.get( "output", "mst", "adda.mst" )
self.mFilenameResult = self.mConfig.get( "output", "result", "adda.result" )
self.mFilenameAlignments = self.mConfig.get("output","align", "adda.align" )
self.mFilenameSegments = self.mConfig.get("output","segments", "adda.segments" )
self.mFilenames = (self.mFilenameSummary,)
def __init__(self, *args, **kwargs):
AddaModuleBlock.__init__(self, *args, **kwargs)
self.mFilenameGraph = self.mConfig.get("files", "output_graph",
"adda.graph")
self.mFilenameDomains = self.mConfig.get("files", "output_domains",
"adda.domains")
self.mEvalueThresholdTrustedLinks = float(
self.mConfig.get("align", "evalue_threshold_trusted_links", -12.0))
self.mFilenameDomainGraph = self.mConfig.get("files",
"output_domaingraph",
"adda.domaingraph.gz")
cadda.setFilenameGraph(self.mFilenameGraph)
cadda.setFilenameDomains(self.mFilenameDomains)
cadda.setLogLevel(self.mLogLevel)
cadda.setEvalueThresholdTrustedLinks(self.mEvalueThresholdTrustedLinks)
self.mFilenames = (self.mFilenameDomainGraph, )
def __init__(self, *args, **kwargs ):
AddaModuleBlock.__init__( self, *args, **kwargs )
self.mFilenameGraph = self.mConfig.get( "files", "output_graph")
self.mFilenameIndex = self.mConfig.get( "files", "output_index")
self.mFilenameTransfers = self.mConfig.get( "files", "output_fit_transfer" )
self.mFilenameFit = self.mConfig.get( "files", "output_fit", "adda.fit" )
self.mFilenameNids = self.mConfig.get( "files", "output_nids" )
self.mFilenameDomains = self.mConfig.get( "files", "output_domains" )
self.mMaxIterations = int( self.mConfig.get( "optimise", "iterations" ) )
self.mResolution = float( self.mConfig.get( "optimise", "resolution" ) )
self.mMinAbsImprovement = float(self.mConfig.get( "optimise", "min_abs_improvement" ))
self.mMinRelImprovement = float(self.mConfig.get( "optimise", "min_rel_improvement" ))
self.mOutputFilenameProgressImprovement = self.mFilenameDomains + "_progress_improvement.png"
self.mOutputFilenameProgressDomains = self.mFilenameDomains + "_progress_domains.png"
self.mOutputFilenameProgressDomainsPerSequence = self.mFilenameDomains + "_progress_domains_per_sequence.png"
self.mNSequences = len(self.mFasta)
self.mFilenames = ( self.mFilenameDomains, )
def finish(self):
self.info( "sequences: %i input, %i output, %i removed, %i duplicates" %\
(self.mInput, self.mOutput, self.mRemoved, self.mDuplicates ) )
AddaModuleBlock.finish(self)
def finish(self):
self.info( "sequences: %i input, %i output, %i removed, %i duplicates" %\
(self.mInput, self.mOutput, self.mRemoved, self.mDuplicates ) )
AddaModuleBlock.finish( self )