CClust is an R package developed to help with clustering categorical datasets. The package can be installed from GitHub using devtools and then loaded in the usual way.
CClust can be installed from the GitHub repository using the devtools package.
Once installed, the package can be loaded as usual.
The package contains the following functions.
read_fastq obtains NGS datasets from a fastq file, extract and return a list of data information: reads, quality scores and dimension of data.
read_fasta obtains NGS datasets from a fasta file, extract and return a list of data information: reads and dimension of data.
kmodes is for clustering categorical datasets without quality information of the data, different from function, it has six different random initialization methods and three k-means algorithms (Lloyd’s; MacQueen’s; Hartigan and Wong’s algorithm) were adapted to do clustering.
khaplotype is for clustering the NGS datasets with quality scores, only random initialization is avaiable, also three k-means algorithms (Lloyd’s; MacQueen’s; Hartigan and Wong’s algorithm) were adapted to do clustering.
ARI is for computing the adjusted rand index given the estimated assignments and the true assiganments.
plot_cluster
Visulize clusters after using the cluster algorithms implemented in the functions. This cluster plot make use of the function \code{dapc} from package \code{adegenet} and function \code{scatter} from package \code{ade4}. The plot is based on observation assignment and discriminant analysis of principal components, therefore, when running the code, it will asks you to choose the number PCs to retain and choose the number discriminant functions to retain.