fastMNN for study 1 (Count)

Here, fastMNN is executed on the Seurat object for Study 1 (Count) data. The parameters and commands are derived from the fastMNN documentation.

Load libraries

library(batchelor)
library(SeuratWrappers)
library(Seurat)
library(dplyr)
library(ggplot2)

source("~/Supplemental_code/Data_integration.R")

Load datasets for Study 1

load("~/Supplemental_code/unCTC_datasets/Zheng_et_al._Data.RData")
load("~/Supplemental_code/unCTC_datasets/Zheng_et_al._metaData.RData")
load("~/Supplemental_code/unCTC_datasets/Velten_et_al._Data.RData")
load("~/Supplemental_code/unCTC_datasets/Velten_et_al._metaData.RData")
load("~/Supplemental_code/unCTC_datasets/Sarioglu_et_al._Data.RData")
load("~/Supplemental_code/unCTC_datasets/Sarioglu_et_al._metaData.RData")
load("~/Supplemental_code/unCTC_datasets/Jordan_et_al._Data.RData")
load("~/Supplemental_code/unCTC_datasets/Jordan_et_al._metaData.RData")
load("~/Supplemental_code/unCTC_datasets/Aceto_et_al._metaData.RData")
load("~/Supplemental_code/unCTC_datasets/Aceto_et_al._Data.RData")
load("~/Supplemental_code/unCTC_datasets/Yu_et_al._metaData.RData")
load("~/Supplemental_code/unCTC_datasets/Yu_et_al._Data.RData")
load("~/Supplemental_code/unCTC_datasets/Ting_et_al._metaData.RData")
load("~/Supplemental_code/unCTC_datasets/Ting_et_al._Data.RData")

Data integration based on common genes

Data2 = Data_integration(data_list=list(Velten_et_al._Data,
                                      Ting_et_al._Data,
                                      Yu_et_al._Data,
                                      Sarioglu_et_al._Data,
                                      Jordan_et_al._Data,
                                      Aceto_et_al._Data,
                                      Zheng_et_al._Data))
data2_metadata = rbind(Velten_et_al._metaData,
                       Ting_et_al._metaData,
                       Yu_et_al._metaData,
                       Sarioglu_et_al._metaData,
                       Jordan_et_al._metaData,
                       Aceto_et_al._metaData,
                       Zheng_et_al._metaData)

Seurat_obj_study1 <- CreateSeuratObject(counts = Data2, project = "Study_1_data",meta.data = data2_metadata)

# normalize and identify variable features for each dataset independently
Seurat_obj_study1 <- NormalizeData(Seurat_obj_study1, normalization.method = "LogNormalize", scale.factor = 10000)

#Identification of highly variable features (feature selection)
Seurat_obj_study1 <- FindVariableFeatures(Seurat_obj_study1, selection.method = "vst", nfeatures = 2000)%>% 
  ScaleData(verbose = FALSE)

Seurat’s fastMNN wrapper

Seurat_obj_study1 <- RunFastMNN(object.list = SplitObject(Seurat_obj_study1, split.by = "DataID"))
Seurat_obj_study1 <- RunUMAP(Seurat_obj_study1, reduction = "mnn", dims = 1:30)
Seurat_obj_study1 <- FindNeighbors(Seurat_obj_study1, reduction = "mnn", dims = 1:30)
Seurat_obj_study1 <- FindClusters(Seurat_obj_study1)

## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
## 
## Number of nodes: 1184
## Number of edges: 50200
## 
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.6826
## Number of communities: 6
## Elapsed time: 0 seconds

Colorkey for Visualization

ColorKeyDataID = c("coral4","darkcyan","steelblue","orangered",
                   "darkolivegreen4","lightsteelblue3","darkorchid4",
                   "darkslategray","salmon3","paleturquoise1",
                   "mediumaquamarine","greenyellow","black",
                   "deepskyblue3","mediumblue","peru","gold",
                   "gray50","hotpink","khaki3","yellow4","lavender",
                   "cornsilk4","orchid4","yellow3", "darkgreen",
                   "skyblue1","khaki4","tan4","firebrick1","pink")

mnn reduction

DimPlot(Seurat_obj_study1,reduction = "mnn", pt.size = 1, group.by = "DataID")+scale_color_manual(values=ColorKeyDataID)

DimPlot(Seurat_obj_study1,reduction = "mnn", pt.size = 1)+scale_color_manual(values=ColorKeyDataID)

UMAP reduction

DimPlot(Seurat_obj_study1,reduction = "umap", pt.size = 1, group.by = "DataID")+scale_color_manual(values=ColorKeyDataID)

DimPlot(Seurat_obj_study1,reduction = "umap", pt.size = 1)+scale_color_manual(values=ColorKeyDataID)

barplot

df_study1 = as.data.frame(Seurat_obj_study1@reductions$umap@cell.embeddings)
df_study1$Cell_type = Seurat_obj_study1@meta.data$Cell_type
df_study1$DataID = Seurat_obj_study1@meta.data$DataID
df_study1$Clusters = Seurat_obj_study1@meta.data$seurat_clusters

ggplot(df_study1, aes(x=Clusters, fill = Cell_type)) + theme_classic()+
       geom_bar(stat="count")+scale_color_manual()+
       scale_fill_manual(values = c("dodgerblue4","firebrick3","darkgreen","dark turquoise"))+
       theme(legend.text = element_text(size=14),
        plot.title = element_text(size=16),
        legend.title=element_text(size=20),axis.text=element_text(size=20),
        axis.title=element_text(size=22,face="bold")) +
        guides(colour = guide_legend(override.aes = list(size = 6)))

ARI, NMI and Cluster purity

aricode::ARI(df_study1$Clusters,df_study1$Cell_type)

## [1] 0.05272533

aricode::NMI(df_study1$Clusters,df_study1$Cell_type)

## [1] 0.03954897

ClusterPurity <- function(clusters, classes) {
  sum(apply(table(classes, clusters), 2, max)) / length(clusters)
}
ClusterPurity(df_study1$Clusters,df_study1$Cell_type)

## [1] 0.8758446