gaucho: Genetic Algorithm for Understanding Clonal Heterogeneity and...

In gaucho: Genetic Algorithms for Understanding Clonal Heterogeneity and Ordering

Description

Use a genetic algorithm to find the relationships between the values in an input file - the package was written to deal with single nucleotide variants (SNVs) in mixtures of cancer cells, but it will work with any mixture. It will calculate appropriate phylogenetic relationships between clones them and the proportion of each clone that each sample is composed of. For detailed usage, please read the accompanying vignette.

Usage

gaucho(observations, number_of_clones, pop_size = 100, mutation_rate = 0.8,
  iterations = 1000, stoppingCriteria = round(iterations/5),
  parthenogenesis = 2, nroot = 0, contamination = 0,
  check_validity = TRUE)

Arguments

observations	Observation data frame where each row represents an SNV and each column represents a discrete sample separated by time or space. Note that the data frame must have column names and row names. Every value must be a proportion between 0 and 1. See details
number_of_clones	An integer number of clones to be considered
pop_size	The number of individuals in each generation
mutation_rate	The likelihood of each individual undergoing mutation per generation
iterations	The maximum number of generations to run
stoppingCriteria	The number of consecutive generations without improvement that will stop the algorithm. Default value is 20% of iterations.
parthenogenesis	The number of best-fitness individuals allowed to survive each generation
nroot	Number of roots the phylogeny is expected to have.When nroot=0, a random integer between 1 and the number of clones is generated for each phylogeny
contamination	Is the input contaminated? If set to 1, an extra clone is created in which to place inferred contaminants
check_validity	Unless set to false, eliminate any clones with no new mutations, disallow those clones. Increases computational overheads.

Details

The input data should be a data.frame containing proportions of cells that contain a feature. There are a number of ways to create these data, including merging the balance of alleles and copy number of an SNV using the equation min(1,r*CN/(r+R)), where CN is the copy number, r is the number of non-reference reads and R is the number of reference reads. For example, if a site were sequenced to a depth of 100x, with 25 non-reference reads and 75 reference reads and diploid copy number, the result would be min(1,25*2/(25+75)) = 0.5. Therefore, 50% of the cells in the sample contain the SNV. Further details are available in the accompanying vignette.

Value

Returns an object of class ga ga-class. Note that the number of clones and number of cases are stored in the unused min and max slots of the output object.

Author(s)

Alex Murison [e-mail] and Christopher Wardell [e-mail]

See Also

ga-class, ga, gauchoReport, gaucho_simple_data, gaucho_hidden_data, gaucho_synth_data, gaucho_synth_data_jittered, BYB1_G07_pruned

Examples

## The vignette provides far more in-depth explanation and examples ##

## Load the included simple example data
gaucho_simple_data = read.table(file.path(system.file("extdata",package="gaucho"),"gaucho_simple_data.txt"),header=TRUE,row.names=1)

## Run gaucho using 3 clones and a phylogeny with a single root
solution=gaucho(gaucho_simple_data, number_of_clones=3,nroot=1,iterations=1000)

## Create the four output plots
gauchoReport(gaucho_simple_data,solution,outType="fitness")
gauchoReport(gaucho_simple_data,solution,outType="heatmap")
gauchoReport(gaucho_simple_data,solution,outType="phylogeny")
gauchoReport(gaucho_simple_data,solution,outType="proportion")

## Output the solution and plots in the current working directory
# gauchoReport(gaucho_simple_data,solution)

gaucho documentation built on May 2, 2018, 2:37 a.m.