,
Duncan Taylor [aut] | Title: | Mixtures of Discrete Laplace Distributions using Numerical Optimisation |
|---|---|
| Description: | Fit a mixture of Discrete Laplace distributions using plain numerical optimisation. This package has similar applications as the 'disclapmix' package that uses an EM algorithm. |
| Authors: | Maarten Kruijver [aut, cre] ,
Duncan Taylor [aut] |
| Maintainer: | Maarten Kruijver <[email protected]> |
| License: | GPL (>= 2) |
| Version: | 0.6.2 |
| Built: | 2024-11-03 03:50:16 UTC |
| Source: | https://github.com/mkruijver/disclapmix2 |
An extension to the *disclapmix* method in the *disclapmix* package that supports duplicated loci and other non-standard haplotypes.
disclapmix2( x, number_of_clusters, include_2_loci = FALSE, remove_non_standard_haplotypes = TRUE, use_stripped_data_for_initial_clustering = FALSE, initial_y_method = "pam", verbose = 0L )disclapmix2( x, number_of_clusters, include_2_loci = FALSE, remove_non_standard_haplotypes = TRUE, use_stripped_data_for_initial_clustering = FALSE, initial_y_method = "pam", verbose = 0L )
x |
DataFrame. Columns should be one character vector for each locus |
number_of_clusters |
The number of clusters to fit the model for. |
include_2_loci |
Should duplicated loci be included or excluded from the analysis? |
remove_non_standard_haplotypes |
Should observations that are not single integer alleles be removed? |
use_stripped_data_for_initial_clustering |
Should non_standard data be removed for the initial clustering? |
initial_y_method |
Which cluster method to use for finding initial central haplotypes, y: pam (recommended) or clara. |
verbose |
Set to 1 (or higher) to print optimisation details. Default is 0. |
List.
require(disclapmix) data(danes) x <- as.matrix(danes[rep(seq_len(nrow(danes)), danes$n), -ncol(danes)]) x2 <- as.data.frame(sapply(danes[rep(seq_len(nrow(danes)), danes$n), -ncol(danes)], as.character)) dlm_fit <- disclapmix(x, clusters = 3L) dlm2_fit <- disclapmix2(x2, number_of_clusters = 3) stopifnot(all.equal(dlm_fit$logL_marginal, dlm2_fit$log_lik))require(disclapmix) data(danes) x <- as.matrix(danes[rep(seq_len(nrow(danes)), danes$n), -ncol(danes)]) x2 <- as.data.frame(sapply(danes[rep(seq_len(nrow(danes)), danes$n), -ncol(danes)], as.character)) dlm_fit <- disclapmix(x, clusters = 3L) dlm2_fit <- disclapmix2(x2, number_of_clusters = 3) stopifnot(all.equal(dlm_fit$logL_marginal, dlm2_fit$log_lik))
Count the number of times each haplotype occurs
haplotype_counts(x)haplotype_counts(x)
x |
DataFrame (by locus) of character vectors containing haplotypes (rows) where alleles are separated by comma's, e.g. "13,14.2" is a haplotype |
Integer vector with count for each row in DataFrame
# read haplotypes h <- readxl::read_excel(system.file("extdata","South_Australia.xlsx", package = "disclapmix2"), col_types = "text")[-c(1,2)] # obtain counts counts <- disclapmix2::haplotype_counts(h) # all haplotypes in the dataset are unique stopifnot(all(counts == 1))# read haplotypes h <- readxl::read_excel(system.file("extdata","South_Australia.xlsx", package = "disclapmix2"), col_types = "text")[-c(1,2)] # obtain counts counts <- disclapmix2::haplotype_counts(h) # all haplotypes in the dataset are unique stopifnot(all(counts == 1))
Compute the profile probability for a new profile that was not used in the original fit.
profile_pr_by_locus_and_cluster(x, fit)profile_pr_by_locus_and_cluster(x, fit)
x |
DataFrame. Columns should be one character vector for each locus |
fit |
Output from disclapmix2 |
Numeric.
require(disclapmix) data(danes) x <- as.data.frame(sapply(danes[rep(seq_len(nrow(danes)), danes$n), -ncol(danes)], as.character)) dlm2_fit <- disclapmix2(x, number_of_clusters = 3) new_profile <- structure(list(DYS19 = "14", DYS389I = "13", DYS389II = "29", DYS390 = "22", DYS391 = "9", DYS392 = "15", DYS393 = "13", DYS437 = "14", DYS438 = "11", DYS439 = "12"), row.names = 1L, class = "data.frame") profile_pr_by_locus_and_cluster(x = new_profile, dlm2_fit)require(disclapmix) data(danes) x <- as.data.frame(sapply(danes[rep(seq_len(nrow(danes)), danes$n), -ncol(danes)], as.character)) dlm2_fit <- disclapmix2(x, number_of_clusters = 3) new_profile <- structure(list(DYS19 = "14", DYS389I = "13", DYS389II = "29", DYS390 = "22", DYS391 = "9", DYS392 = "15", DYS393 = "13", DYS437 = "14", DYS438 = "11", DYS439 = "12"), row.names = 1L, class = "data.frame") profile_pr_by_locus_and_cluster(x = new_profile, dlm2_fit)
List unique haplotypes with their counts
unique_haplotype_counts(x)unique_haplotype_counts(x)
x |
DataFrame (by locus) of character vectors containing haplotypes (rows) where alleles are separated by comma's, e.g. "13,14.2" is a haplotype |
DataFrame with unique rows and a Count column added at the end
# read haplotypes h <- readxl::read_excel(system.file("extdata","South_Australia.xlsx", package = "disclapmix2"), col_types = "text")[-c(1,2)] # obtain counts unique_counts <- disclapmix2::unique_haplotype_counts(h) # all haplotypes in the dataset are unique stopifnot(all(unique_counts$Count == 1))# read haplotypes h <- readxl::read_excel(system.file("extdata","South_Australia.xlsx", package = "disclapmix2"), col_types = "text")[-c(1,2)] # obtain counts unique_counts <- disclapmix2::unique_haplotype_counts(h) # all haplotypes in the dataset are unique stopifnot(all(unique_counts$Count == 1))