Multiple-Integrating: Integrating Multiple Datasets

In bioinfoDZ/RISC: Robust Integration of Single-Cell RNA-Seq Datasets

Integrating Multiple Datasets

Description

The "scMultiIntegrate" function can be used for data integration of multiple datasets, it is basically based on our new approach RPCI (reference principal components integration), which decomposes all the target datasets based on the reference data. The output of this function is RISC object, including the integrated eigenvectors and aligned gene expression values.

Usage

scMultiIntegrate(
  objects,
  eigens = 10,
  add.Id = NULL,
  var.gene = NULL,
  align = "OLS",
  npc = 50,
  adjust = TRUE,
  ncore = 1,
  seed = 123
)

Arguments

objects	The list of multiple RISC objects: listobject1, object2, object3, .... The first set is the reference to generate gene-eigenvectors.
eigens	The number of eigenvectors used for data integration.
add.Id	Add a vector of Id to label different datasets, a character vector.
var.gene	Define the variable genes manually. Here input a vector of gene names as variable genes
align	The method for alignment of gene expression values: "Optimal" for alignment by experience, "Predict" for alignment by RPCI prediction, and "OLS" for alignment by the ordinary linear regression.
npc	The number of the PCs returns from "scMultiIntegrate" function, they are usually used for the subsequent analyses, like cell embedding and cell clustering.
adjust	Whether adjust the number of eigenvectors.
ncore	The number of multiple cores for data integration.
seed	The random seed to keep consistent result.

References

Liu et al., Nature Biotech. (2021)

Examples

obj1 = raw.mat[[3]]
obj2 = raw.mat[[4]]
obj0 = list(obj1, obj2)
var0 = intersect(obj1@vargene, obj2@vargene)
obj0 = scMultiIntegrate(obj0, eigens = 8, var.gene = var0, align = 'Predict', 
                        npc = 20, add.Id = c("Set1", "Set2"), ncore = 2)
obj0 = scUMAP(obj0, npc = 8, use = "PLS", dist = 0.001, neighbors = 15)
DimPlot(obj0, slot = "cell.umap", colFactor = "Set", size = 2)
DimPlot(obj0, slot = "cell.umap", colFactor = "Group", size = 2, label = TRUE)

bioinfoDZ/RISC documentation built on March 30, 2024, 9:19 p.m.