samleenz/ncla

ncla

ncla is an R package written to help with mapping samples between different datasets using transcriptomic data on the basis of measuring dissimilarity

Installation

You can install the development version of ncla from GitHub with:

# install.packages("devtools")
devtools::install_github("samleenz/ncla")

Citing

See the manuscript:

ncla's Issues

induce bipartite neigbour graph

Given a set of query samples X, return a network diagram showing the mapping to the nearest neghbours of the other type. Similar to the bespoke analysis I did for the PDX analysis but for the general case (would likely not have the dual edge type)

look into using the ggraph package for this

xdissim

add check that the matrices have same features

require philentropy version 0.5?

>     di <- sapply(2:nrow(xyMat), function(i){
+       ncla::dissim(
+         xyMat[c(1, i), ],
+         d
+       )[1,2] # [1,2] is to select the top-right of the 2x2 distance matrix
+     })
Error in philentropy::distance(x, "euclidean", mute.message = TRUE) :
  unused argument (mute.message = TRUE)

error when using philentropy version 0.4, fixed on update to latest philentropy version

typo in subsetGenes documentation

"Halmarks"

fn: transform distance matrix into tidy table

write a utility function that takes a square distance matrix, and returns the upper-triangle of it as a 3 column tibble. add argument to determine whether the diagonal of the matrix will be returned.

X	Y	distance
a	b	0.2

Add SE wrapper for map_samples

map_samples_se <- function(query, reference, f, i_q, i_r){
  q <- assay(query, i_q) |>
    t()
  
  r <- assay(reference, i_r) |>
    t()
  
  res <- ncla::map_samples(q, r, f)
  
  res
}

• write documentation
• add + update

bug in plotGG_PCAscree_eda

stringr func not declared properly

Error in `dplyr::mutate()`:
! Problem while computing `PC_int = as.numeric(str_remove(PC,
  "PC"))`.
Caused by error in `str_remove()`:
! could not find function "str_remove"
Run `rlang::last_error()` to see where the error occurred.

add shape parameter to plotGG functions

is there a flexible way to do it?, work out and do it. annoying not to have

ncla()

Given two datasets return k-NN for each sample in dataset one from dataset two.

This should probably be done in it's own branch?