# This is known to work at least under the following conditions: # - Oxford Nanopore Ultra-Long (UL) reads with read N50 50 Kb or more. # - Guppy 5 base caller or newer with "super" accuracy. # - Human genome. # - Coverage 60x to 80x. If you have more coverage, # adjust --Reads.minReadLength or --Reads.desiredCoverage # to bring coverage down to this range. # For a human genome you can set --Reads.desiredCoverage 200G. # Under the above conditions, this should give an assembly with # N50 in the tens of Mb. [Reads] minReadLength = 50000 noCache = True [Kmers] k = 14 [MinHash] minBucketSize = 10 maxBucketSize = 50 minFrequency = 5 [Align] alignMethod = 3 downsamplingFactor = 0.05 matchScore = 6 sameChannelReadAlignment.suppressDeltaThreshold = 30 # The following Align parameters are set to very permissive values to allow the majority of alignments # to be assessed during the initial stage of automatic alignment parameter selection # (ReadGraph.creationMethod 2). maxSkip = 100 maxDrift = 100 maxTrim = 100 minAlignedMarkerCount = 10 minAlignedFraction = 0.1 [ReadGraph] # This uses the observed distribution of alignment statistics to choose thresholds for # maxSkip, maxDrift, maxTrim, minAlignedMarkerCount, and minAlignedFraction creationMethod = 2 maxAlignmentCount = 12 strandSeparationMethod = 2 [MarkerGraph] simplifyMaxLength = 10,100,1000,10000,100000 crossEdgeCoverageThreshold = 3 # Adaptive estimation of coverage threshold to generate marker graph vertices. minCoverage = 0 [Assembly] consensusCaller = Bayesian:guppy-5.0.7-b detangleMethod = 2