Up: Component summary Component


Reclustering tool for PyClone variant clustering. Work in progress. This tool supplants the PyClone analysis and requires PyClone version 0.13. For more information about PyClone, see https://bitbucket.org/aroth85/pyclone/wiki/Home and Roth et al., Nat Methods 11: 396, 2014.

Version 1.0
Bundle sequencing
Categories Analysis
Authors Antti Hakkinen (antti.e.hakkinen@helsinki.fi)
Issue tracker View/Report issues
Requires Python ; PyClone
Source files component.xml launcher.py
Usage Example with default values


Name Type Mandatory Description
config YAML Mandatory PyClone configuration file
trace BinaryFolder Optional PyClone trace directory (overrides that specified in the configuration file)
mutationFiles Array<YAML> Optional Mutation files in the YAML format


Name Type Description
clusteringResults CSVList Results of clustering
plots BinaryFolder Output directory for the plots


Name Type Default Description
burnin int 0 Number of MCMC samples to discard from the beginning. Prior to convergence, the MCMC series features an initial transient which is controlled by the initial parameters and not the data. This initial transient should be specified such that it can be discarded from the subsequent analysis.
clusterPValues boolean false If enabled, computes p-values for each locus to belong into its assigned cluster. The values are output in the loci table.
infCoverage float 1e100 Variants with coverage no smaller than the specified will be considered as known, and will be spliced off of clusters with unequal values.
meshSize int 101 Number of mesh points for density estimation. The default of 101 allows subdivision of 1% in cellular prevalences.
minAssignLr float 0 Minimum likelihood ratio to assign into a cluster. A value of 0 will assign all variants to a cluster, while a larger value eliminates uncertain assignments.
minClusterSize int 1 Minimum number of variants in a cluster.
minCreateLr float 0 Minimum likelihood ratio to create a cluster. A value of 0 will keep all clusters, while for a larger value clusters with low evidence of cluster linkage are rejected.
plotType string "" Comma separated list of the requested plot types. Options: "similarity_matrix", "filter_metrics", "cluster_density".
thin int 1 Thinning ratio for the MCMC series in samples. In an MCMC series, the consecutive samples are correlated. These correlations will skew higher-order statistics such as variance estimates. This can be mitigated by specifying T such that each T-sample string is thinned into a single sample.
trim float 0 Fraction of data to be trimmed from the summary statistics. A value of 0 produces regular mean/std, while a larger value excludes a specified fraction of outliers.

Test cases

Test case Parameters IN
default properties config trace mutationFiles clusteringResults plots

minCreateLr = 0.5,
minAssignLr = 0.5,
trim = .75,
clusterPValues = true

Generated 2019-02-08 07:42:12 by Anduril 2.0.0