Find Alternative expression

Source code

Description

Using linear regression model (implemented by DEXSeq) or PermTest method to calculate delta ratio and simulate p value for each event.

Usage

FindAltExp(
  object = NULL,
  cells.1 = NULL,
  cells.2 = NULL,
  ident.1 = NULL,
  ident.2 = NULL,
  assay = NULL,
  bind.name = "gene_name",
  test.use = c("DEXSeq", "PermTest"),
  bind.assay = NULL,
  features = NULL,
  bind.features = NULL,
  min.cells = 10,
  return.thresh = NULL,
  mode = c(1, 2, 3),
  threads = 0,
  perm = 100,
  seed = 999,
  pseudo.group = 3,
  debug = FALSE
)

Arguments

object A Seurat object.
cells.1 Vector of cell names belong to group 1. Conflict with ident.1
cells.2 Vector of cell names for comparsion. Conflict with ident.2
ident.1 Identify class to test, if not set will compare all groups one by one
ident.2 A second class for comparsion. If NULL (default), use all other cells for comparison.
assay Test assay (X). Default assay will be used if not set.
bind.name Title name for binding features in the meta table. Consider most users start Yano to perform alternative splicing analysis, the default bind.name set to “gene_name”.
bind.assay Bind assay (Y). If not set, will aggregate all X values of the same block.
features Candidate list to test. If not set, will use AutoCorrFeatures(object, assay = assay).
bind.features Candidate list for bind features to test. If not set, will test all covered.
min.cells Used to filter candiate features or binding features. Require them at least expressed in min.cells. Default is 10.
return.thresh Only return markers that have a p-value < return.thresh. Default is NULL.
mode Test mode. For mode 1, X (test feature) vs Y (binding feature). For mode 2, X vs (Y-X). For mode 3, X vs (Y+X).
threads Threads. For DEXSeq, threads will set to 1. For other methods, threads set to 0, which will auto check the CPU cores and set threads = number of CPU cores -1.
perm Permutation steps for calculate statistical of delta-ratio. Default is 100.
seed Seed for generate random number. Default is 999.
debug Print debug logs. Will auto set thread to 1. Default is FALSE.
pesudo.group Aggregate counts into groups for each clusters. Used only for DEXSeq.

Details

This function first aggregates all raw counts per feature for each group and then perform alternative expression analysis with DEXSeq or PermTest method. For DEXSeq, counts per group will divide into pesudo-groups first. For PermTest method, *deltarati**o = Xa/Ya − Xb/Yb*. It then employs a permutation method to randomize the cells in the two groups 100 times (by default) to evaluate the mean and standard deviation of delta-ratio. A p-value is calculated using a t-test. The PermTest method is not well tested. Therefore, we use DEXSeq method in default, but it’s very slow if input a lot of features.

Back to top