corralm_matlist | R Documentation |

This multi-table adaptation of correpondence analysis applies the same scaling technique and enables data alignment by finding a set of embeddings for each dataset within shared latent space.

corralm_matlist( matlist, method = c("irl", "svd"), ncomp = 30, rtype = c("indexed", "standardized", "hellinger", "freemantukey", "pearson"), vst_mth = c("none", "sqrt", "freemantukey", "anscombe"), rw_contrib = NULL, ... ) corralm_sce( sce, splitby, method = c("irl", "svd"), ncomp = 30, whichmat = "counts", fullout = FALSE, rw_contrib = NULL, ... ) corralm(inp, whichmat = "counts", fullout = FALSE, ...) ## S3 method for class 'corralm' print(x, ...)

`matlist` |
(for |

`method` |
character, the algorithm to be used for svd. Default is irl. Currently supports 'irl' for irlba::irlba or 'svd' for stats::svd |

`ncomp` |
numeric, number of components; Default is 30 |

`rtype` |
character indicating what type of residual should be computed; options are '"indexed"', '"standardized"' (or '"pearson"' is equivalent), '"freemantukey"', and '"hellinger"'; defaults to '"standardized"' for |

`vst_mth` |
character indicating whether a variance-stabilizing transform should be applied prior to calculating chi-squared residuals; defaults to '"none"' |

`rw_contrib` |
numeric vector, same length as the matlist. Indicates the weight that each dataset should contribute to the row weights. When set to NULL the row weights are *not* combined and each matrix is scaled independently (i.e., using their observed row weights, respectively). When set to a vector of all the same values, this is equivalent to taking the mean. Another option is to the number of observations per matrix to create a weighted mean. Regardless of input scale, row weights for each table must sum to 1 and thus are scaled. When this option is specified (i.e., not 'NULL'), the 'rtype' argument will automatically be set to 'standardized', and whatever argument is given will be ignored. |

`...` |
(additional arguments for methods) |

`sce` |
(for |

`splitby` |
character; name of the attribute from |

`whichmat` |
char, when using SingleCellExperiment or other SummarizedExperiment, can be specified. default is 'counts'. |

`fullout` |
boolean; whether the function will return the full |

`inp` |
list of matrices (any type), a |

`x` |
(print method) corralm object; the list output from |

`corralm`

is a wrapper for `corralm_matlist`

and `corralm_sce`

, and can be called on any of the acceptable input types (see `inp`

below).

When run on a list of matrices, a list with the correspondence analysis matrix decomposition result, with indices corresponding to the concatenated matrices (in order of the list):

`d`

a vector of the diagonal singular values of the input

`mat`

(from SVD output)`u`

a matrix of with the left singular vectors of

`mat`

in the columns (from SVD output)`v`

a matrix of with the right singular vectors of

`mat`

in the columns. When cells are in the columns, these are the cell embeddings. (from SVD output)`eigsum`

sum of the eigenvalues for calculating percent variance explained

For SingleCellExperiment input, returns the SCE with embeddings in the reducedDim slot 'corralm'

For a list of `SingleCellExperiment`

s, returns a list of the SCEs with the embeddings in the respective `reducedDim`

slot 'corralm'

.

listofmats <- list(matrix(sample(seq(0,20,1),1000,replace = TRUE),nrow = 25), matrix(sample(seq(0,20,1),1000,replace = TRUE),nrow = 25)) result <- corralm_matlist(listofmats) library(DuoClustering2018) library(SingleCellExperiment) sce <- sce_full_Zhengmix4eq()[1:100,sample(1:3500,100,replace = FALSE)] colData(sce)$Method <- matrix(sample(c('Method1','Method2'),100,replace = TRUE)) result <- corralm_sce(sce, splitby = 'Method') listofmats <- list(matrix(sample(seq(0,20,1),1000,replace = TRUE),nrow = 20), matrix(sample(seq(0,20,1),1000,replace = TRUE),nrow = 20)) corralm(listofmats) library(DuoClustering2018) library(SingleCellExperiment) sce <- sce_full_Zhengmix4eq()[seq(1,100,1),sample(seq(1,3500,1),100,replace = FALSE)] colData(sce)$Method <- matrix(sample(c('Method1','Method2'),100,replace = TRUE)) result <- corralm(sce, splitby = 'Method') # default print method for corralm objects

Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.