Validation of scSpace on human embryonic heart scRNA-seq data
library(Seurat)
## Attaching SeuratObject
## Attaching sp
library(ggplot2)
library(ggsignif)
library(ggforce)
library(mclust)
## Package 'mclust' version 6.0.0
## Type 'citation("mclust")' for citing this R package in publications.
library(dplyr)
##
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
##
## filter, lag
## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union
library(cowplot)
library(reshape2)
library(ggExtra)
library(NMI)
Load data
# Ground Truth
load('../data/heart/heart.RDS')
Load custom functions
source('../scripts/utils.R')
Set output dir
output_dir <- '../output/figure3'
if(!dir.exists(output_dir)){
dir.create(output_dir)
} else {
print('Dir already exists!')
}
## [1] "Dir already exists!"
Pre-processing
sc_meta$pseudo_space1 <- pseudo_space[, 1]
sc_meta$pseudo_space2 <- pseudo_space[, 2]
sc_meta$scSpace <- scSpace_cluster$scSpace
color <- c('#3B82F6', '#5EEAD4', '#EAB308', '#10B981', '#854D0E',
'#84CC16', '#881337', '#991B1B', '#EF4444', '#D946EF',
'#F97316', '#A5B4FC', '#0891B2', '#8B5CF6')
Pseudo space of all celltypes
p <- ggplot() +
geom_point(sc_meta, mapping = aes(pseudo_space1, pseudo_space2, color=celltype), size=2)+
scale_color_manual(values = c('#EF4444', '#3B82F6', '#0891B2', '#881337', '#991B1B',
'#EAB308', '#84CC16', '#854D0E', '#D946EF', '#5EEAD4',
'#8B5CF6', '#A5B4FC', '#F97316', '#10B981')) +
theme_classic() +
ggtitle('Pseudo space of all celltypes') +
theme(plot.title = element_text(hjust = 0.5))
p
Pseudo space of three cardiomyocytes
# 3 cardiomyocytes
card_ct <- c('Atrial cardiomyocytes', 'Myoz2-enriched cardiomyocytes', 'Ventricular cardiomyocytes')
sc_meta$celltype2 <- sc_meta$celltype
sc_meta[!sc_meta$celltype %in% card_ct, ]$celltype2 <- 'Others'
p1 <- ggplot() +
geom_point(sc_meta[sc_meta$celltype2 == 'Others', ], mapping = aes(pseudo_space1,pseudo_space2, color = celltype2), size = 2) +
geom_point(sc_meta[sc_meta$celltype2 != 'Others', ], mapping = aes(pseudo_space1,pseudo_space2, color = celltype2), size = 2) +
scale_color_manual(values = c('#EF4444', '#A5B4FC', 'grey90', '#10B981')) +
theme_classic() +
ggtitle('Pseudo space of 3 cardiomyocytes') +
theme(plot.title = element_text(hjust = 0.5))
p1
p2 <- ggplot() +
geom_point(sc_meta[sc_meta$celltype2 != 'Others', ], mapping = aes(pseudo_space1,pseudo_space2, color = celltype2), size = 2) +
scale_color_manual(values = c('#EF4444', '#A5B4FC', '#10B981')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle('Pseudo space of 3 cardiomyocytes') +
theme(plot.title = element_text(hjust = 0.5))
p2 <- ggExtra::ggMarginal(p = p2, type = "density", groupFill = T)
p2
dist_to_atrial <- cal_dist(
meta = sc_meta[sc_meta$celltype %in% card_ct, ],
coord_index = c('pseudo_space1', 'pseudo_space2'),
group_by = 'celltype',
selected_type = 'Atrial cardiomyocytes',
ignore_select_type = TRUE
)
## Beginning normalized distance calculating...
## [1] "Normalized distance calculating done."
dist_to_atrial[dist_to_atrial$group == 'Atrial cardiomyocytes_Myoz2-enriched cardiomyocytes', ]$group <- 'Myoz2-enriched'
dist_to_atrial[dist_to_atrial$group == 'Atrial cardiomyocytes_Ventricular cardiomyocytes', ]$group <- 'Ventricular'
dist_to_ventricular <- cal_dist(
meta = sc_meta[sc_meta$celltype %in% card_ct, ],
coord_index = c('pseudo_space1', 'pseudo_space2'),
group_by = 'celltype',
selected_type = 'Ventricular cardiomyocytes',
ignore_select_type = TRUE
)
## Beginning normalized distance calculating...
## [1] "Normalized distance calculating done."
dist_to_ventricular[dist_to_ventricular$group == 'Ventricular cardiomyocytes_Myoz2-enriched cardiomyocytes', ]$group <- 'Myoz2-enriched'
dist_to_ventricular[dist_to_ventricular$group == 'Ventricular cardiomyocytes_Atrial cardiomyocytes', ]$group <- 'Atrial'
p1 <- ggplot(data=dist_to_atrial, aes(x=group, y=dist, fill=group)) +
geom_boxplot(outlier.shape = NA) +
scale_fill_manual(values = c('#A5B4FC', '#10B981') )+
theme_classic() +
ggtitle('Distance to Atrial cardiomyocytes') +
theme(plot.title = element_text(hjust = 0.5)) +
ggsignif::geom_signif(comparisons = list(c('Myoz2-enriched', 'Ventricular')), test = 'wilcox.test')
dist_to_ventricular$group <- factor(dist_to_ventricular$group, levels = c('Myoz2-enriched', 'Atrial'))
p2 <- ggplot(data=dist_to_ventricular, aes(x=group, y=dist, fill=group)) +
geom_boxplot(outlier.shape = NA) +
scale_fill_manual(values = c('#A5B4FC', '#EF4444') )+
theme_classic() +
ggtitle('Distance to Ventricular cardiomyocytes') +
theme(plot.title = element_text(hjust = 0.5)) +
ggsignif::geom_signif(comparisons = list(c('Myoz2-enriched', 'Atrial')), test = 'wilcox.test')
cowplot::plot_grid(p1, p2)
Spatial-informed clustering
seu_obj <- run_Seurat(sc_meta = sc_meta, sc_data = sc_data, res = 0.44)
## Warning: Feature names cannot have underscores ('_'), replacing with dashes
## ('-')
## Computing nearest neighbor graph
## Computing SNN
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.9326
## Number of communities: 14
## Elapsed time: 0 seconds
sc_meta <- seu_obj@meta.data
sc_meta$tsne1 <- seu_obj@reductions$tsne@cell.embeddings[, 1]
sc_meta$tsne2 <- seu_obj@reductions$tsne@cell.embeddings[, 2]
sc_meta$scSpace <- as.factor(sc_meta$scSpace)
# ground truth
p <- ggplot() +
geom_point(sc_meta, mapping = aes(tsne1, tsne2, color = celltype)) +
geom_circle(aes(x0=-7, y0=33, r=20), color="black", linetype=2) +
scale_color_manual(values = c('#EF4444', '#3B82F6', '#0891B2', '#881337', '#991B1B',
'#EAB308', '#84CC16', '#854D0E', '#D946EF', '#5EEAD4',
'#8B5CF6', '#A5B4FC', '#F97316', '#10B981')) +
theme_classic() +
ggtitle('t-SNE of Ground truth') +
theme(plot.title = element_text(hjust = 0.5))
p
## Warning: Using the `size` aesthetic in this geom was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` in the `default_aes` field and elsewhere instead.
# ARI for all cell types
scspace_ari <- round(adjustedRandIndex(sc_meta$scSpace, sc_meta$celltype), 3)
seurat_ari <- round(adjustedRandIndex(sc_meta$seurat_clusters, sc_meta$celltype), 3)
# scSpace
p1 <- ggplot() +
geom_point(sc_meta, mapping = aes(tsne1, tsne2, color = scSpace)) +
geom_circle(aes(x0=-7, y0=33, r=20), color="black", linetype=2) +
scale_color_manual(values = c('#3B82F6', '#5EEAD4', '#EAB308', '#10B981', '#854D0E',
'#84CC16', '#881337', '#991B1B', '#EF4444', '#D946EF',
'#F97316', '#A5B4FC', '#0891B2', '#8B5CF6')) +
theme_classic() +
ggtitle(paste0('scSpace (ARI: ', scspace_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
# Seurat
p2 <- ggplot() +
geom_point(sc_meta, mapping = aes(tsne1, tsne2, color = seurat_clusters)) +
geom_circle(aes(x0=-7, y0=33, r=20), color="black", linetype=2) +
scale_color_manual(values = c('#EAB308', '#3B82F6', '#10B981', '#5EEAD4', '#854D0E',
'#F97316', '#A5B4FC', '#84CC16', '#881337', '#991B1B',
'#D946EF', '#0891B2', '#8B5CF6', '#EF4444')) +
theme_classic() +
ggtitle(paste0('Seurat (ARI: ', seurat_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
cowplot::plot_grid(p1, p2)
Spatial-informed clustering for three cardiomyocytes
# extract targeted ROI from t-SNE coordinates
card_meta <- sc_meta[sc_meta$tsne1 >= (-7 - 20) & sc_meta$tsne1 < (-7 + 20)
& sc_meta$tsne2 >= (33 - 20) & sc_meta$tsne2 < (33 + 20), ]
# ARI for all cell types
scspace_ari <- round(adjustedRandIndex(card_meta$scSpace, card_meta$celltype), 3)
seurat_ari <- round(adjustedRandIndex(card_meta$seurat_clusters, card_meta$celltype), 3)
# Ground truth
p1 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = celltype)) +
scale_color_manual(values = c('#EF4444', '#A5B4FC', '#10B981')) +
theme_classic() +
ggtitle(paste0('Ground truth')) +
theme(plot.title = element_text(hjust = 0.5))
# scSpace
p2 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = scSpace)) +
scale_color_manual(values = c('#10B981', '#EF4444', '#A5B4FC')) +
theme_classic() +
ggtitle(paste0('scSpace (ARI: ', scspace_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
# Seurat
p3 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = seurat_clusters)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444')) +
theme_classic() +
ggtitle(paste0('Seurat (ARI: ', seurat_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
cowplot::plot_grid(p2, p3, p1)
# ARI/NMI
df <- data.frame(
value = c(adjustedRandIndex(card_meta$scSpace,card_meta$celltype),
adjustedRandIndex(card_meta$seurat_clusters,card_meta$celltype),
NMI(data.frame(cell = 1:nrow(card_meta), clu = card_meta$celltype),
data.frame(cell = 1:nrow(card_meta), clu = card_meta$scSpace))$value,
NMI(data.frame(cell = 1:nrow(card_meta), clu = card_meta$celltype),
data.frame(cell = 1:nrow(card_meta), clu = card_meta$seurat_clusters))$value),
method = c('scSpace', 'Seurat', 'scSpace', 'Seurat'),
index = c('ARI', 'ARI', 'NMI', 'NMI'))
write.csv(df, file = paste0(output_dir, '/Clustering_result_for_cardiomyocytes.csv'))
p <- ggplot(df, aes(index, value, fill = method, color = method)) +
geom_bar(stat = 'identity', position=position_dodge(width = 0.9), width = 0.75) +
geom_text(aes(label = round(value, 3)), position=position_dodge(width = 0.9), vjust=-1) +
scale_fill_manual(values = c('#984EA3','#4DAF4A')) +
scale_color_manual(values = c('black', 'black')) +
theme_classic() +
ggtitle(paste0('Clustering ARI / NMI for cardiomyocytes')) +
theme(plot.title = element_text(hjust = 0.5))
p
Pseudo space of three clusters of cardiomyocytes defined by
scSpace
sc_meta$celltype3 <- sc_meta$scSpace
sc_meta$celltype3 <- as.character(sc_meta$celltype3)
sc_meta[setdiff(rownames(sc_meta), rownames(card_meta)), ]$celltype3 <- 'Others'
p <- ggplot() +
geom_point(sc_meta[sc_meta$celltype3 == 'Others', ], mapping = aes(pseudo_space1,pseudo_space2, color = celltype3), size = 2) +
geom_point(sc_meta[sc_meta$celltype3 != 'Others', ], mapping = aes(pseudo_space1,pseudo_space2, color = celltype3), size = 2) +
scale_color_manual(values = c('#A5B4FC', '#10B981', '#EF4444', 'grey90')) +
theme_classic() +
ggtitle('Pseudo space of 3 clusters of cardiomyocytes defined by scSpace') +
theme(plot.title = element_text(hjust = 0.5))
p
card_meta$scSpace <- as.character(card_meta$scSpace)
dist_to_c4 <- cal_dist(
meta = card_meta,
coord_index = c('pseudo_space1', 'pseudo_space2'),
group_by = 'scSpace',
selected_type = '3',
ignore_select_type = TRUE
)
## Beginning normalized distance calculating...
## [1] "Normalized distance calculating done."
dist_to_c4[dist_to_c4$group == '3_11', ]$group <- 'C12'
dist_to_c4[dist_to_c4$group == '3_8', ]$group <- 'C9'
dist_to_c9 <- cal_dist(
meta = card_meta,
coord_index = c('pseudo_space1', 'pseudo_space2'),
group_by = 'scSpace',
selected_type = '8',
ignore_select_type = TRUE
)
## Beginning normalized distance calculating...
## [1] "Normalized distance calculating done."
dist_to_c9[dist_to_c9$group == '8_3', ]$group <- 'C4'
dist_to_c9[dist_to_c9$group == '8_11', ]$group <- 'C12'
p1 <- ggplot(data=dist_to_c4, aes(x=group, y=dist, fill=group)) +
geom_boxplot(outlier.shape = NA) +
scale_fill_manual(values = c('#A5B4FC', '#EF4444') )+
theme_classic() +
ggtitle('Distance to C4') +
theme(plot.title = element_text(hjust = 0.5)) +
ggsignif::geom_signif(comparisons = list(c('C12', 'C9')), test = 'wilcox.test')
p2 <- ggplot(data=dist_to_c9, aes(x=group, y=dist, fill=group)) +
geom_boxplot(outlier.shape = NA) +
scale_fill_manual(values = c('#A5B4FC', '#10B981') )+
theme_classic() +
ggtitle('Distance to C9') +
theme(plot.title = element_text(hjust = 0.5)) +
ggsignif::geom_signif(comparisons = list(c('C12', 'C4')), test = 'wilcox.test')
cowplot::plot_grid(p1, p2)
Expression patterns of marker genes of three clusters of
cardiomyocytes defined by scSpace in pseudo space
Idents(seu_obj) <- 'scSpace'
markers <- FindAllMarkers(seu_obj, only.pos = TRUE, logfc.threshold = 0.5)
## Calculating cluster 0
## Calculating cluster 1
## Calculating cluster 2
## Calculating cluster 3
## Calculating cluster 4
## Calculating cluster 5
## Calculating cluster 6
## Calculating cluster 7
## Calculating cluster 8
## Calculating cluster 9
## Calculating cluster 10
## Calculating cluster 11
## Calculating cluster 12
## Calculating cluster 13
markers <- markers[markers$p_val_adj < 0.05, ]
markers <- markers[markers$cluster %in% c(3, 8, 11), ]
write.csv(markers, file = paste0(output_dir, '/DEGs_for_three_clusters_scSpace.csv'))
# C4
genes <- c('MYH7','SORBS2','NAV1','TGM2','FBN2')
sc_meta$gene1 <- as.numeric(seu_obj@assays$RNA@data[genes[1], ])
p1 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(tsne1, tsne2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene1 == 0, , ],mapping = aes(tsne1, tsne2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene1>0, , ],mapping = aes(tsne1, tsne2, color=gene1)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[1], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
p2 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene1 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene1>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[1], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene2 <- as.numeric(seu_obj@assays$RNA@data[genes[2], ])
p3 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(tsne1, tsne2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene2 == 0, , ],mapping = aes(tsne1, tsne2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene2>0, , ],mapping = aes(tsne1, tsne2, color=gene2)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[2], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
p4 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene2 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene2>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[2], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene3 <- as.numeric(seu_obj@assays$RNA@data[genes[3], ])
p5 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(tsne1, tsne2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene3 == 0, , ],mapping = aes(tsne1, tsne2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene3>0, , ],mapping = aes(tsne1, tsne2, color=gene3)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[3], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
p6 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene3 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene3>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[3], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene4 <- as.numeric(seu_obj@assays$RNA@data[genes[4], ])
p7 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(tsne1, tsne2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene4 == 0, , ],mapping = aes(tsne1, tsne2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene4>0, , ],mapping = aes(tsne1, tsne2, color=gene4)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[4], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
p8 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene4 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene4>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[4], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene5 <- as.numeric(seu_obj@assays$RNA@data[genes[5], ])
p9 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(tsne1, tsne2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene5 == 0, , ],mapping = aes(tsne1, tsne2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene5>0, , ],mapping = aes(tsne1, tsne2, color=gene5)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[5], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
p10 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '3', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene5 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '3'& sc_meta$gene5>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[5], ' (C4)')) +
theme(plot.title = element_text(hjust = 0.5))
cowplot::plot_grid(p1, p3, p5, p7, p9, p2, p4, p6, p8, p10, ncol = 5)
# C9
genes <- c('MYH6','NR2F1','NR2F2','ATP2A2','CPNE5')
sc_meta$gene1 <- as.numeric(seu_obj@assays$RNA@data[genes[1], ])
p1 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(tsne1, tsne2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene1 == 0, , ],mapping = aes(tsne1, tsne2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene1>0, , ],mapping = aes(tsne1, tsne2, color=gene1)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[1], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
p2 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene1 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene1>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[1], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene2 <- as.numeric(seu_obj@assays$RNA@data[genes[2], ])
p3 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(tsne1, tsne2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene2 == 0, , ],mapping = aes(tsne1, tsne2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene2>0, , ],mapping = aes(tsne1, tsne2, color=gene2)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[2], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
p4 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene2 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene2>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[2], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene3 <- as.numeric(seu_obj@assays$RNA@data[genes[3], ])
p5 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(tsne1, tsne2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene3 == 0, , ],mapping = aes(tsne1, tsne2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene3>0, , ],mapping = aes(tsne1, tsne2, color=gene3)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[3], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
p6 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene3 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene3>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[3], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene4 <- as.numeric(seu_obj@assays$RNA@data[genes[4], ])
p7 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(tsne1, tsne2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene4 == 0, , ],mapping = aes(tsne1, tsne2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene4>0, , ],mapping = aes(tsne1, tsne2, color=gene4)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[4], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
p8 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene4 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene4>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[4], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene5 <- as.numeric(seu_obj@assays$RNA@data[genes[5], ])
p9 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(tsne1, tsne2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene5 == 0, , ],mapping = aes(tsne1, tsne2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene5>0, , ],mapping = aes(tsne1, tsne2, color=gene5)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[5], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
p10 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '8', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene5 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '8'& sc_meta$gene5>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[5], ' (C9)')) +
theme(plot.title = element_text(hjust = 0.5))
cowplot::plot_grid(p1, p3, p5, p7, p9, p2, p4, p6, p8, p10, ncol = 5)
# C12
genes <- c('MYOZ2','SMPX','FABP3','FITM1','PLN')
sc_meta$gene1 <- as.numeric(seu_obj@assays$RNA@data[genes[1], ])
p1 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(tsne1, tsne2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene1 == 0, , ],mapping = aes(tsne1, tsne2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene1>0, , ],mapping = aes(tsne1, tsne2, color=gene1)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[1], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
p2 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene1 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene1>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene1)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[1], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene2 <- as.numeric(seu_obj@assays$RNA@data[genes[2], ])
p3 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(tsne1, tsne2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene2 == 0, , ],mapping = aes(tsne1, tsne2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene2>0, , ],mapping = aes(tsne1, tsne2, color=gene2)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[2], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
p4 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene2 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene2>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene2)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[2], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene3 <- as.numeric(seu_obj@assays$RNA@data[genes[3], ])
p5 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(tsne1, tsne2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene3 == 0, , ],mapping = aes(tsne1, tsne2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene3>0, , ],mapping = aes(tsne1, tsne2, color=gene3)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[3], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
p6 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene3 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene3>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene3)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[3], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene4 <- as.numeric(seu_obj@assays$RNA@data[genes[4], ])
p7 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(tsne1, tsne2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene4 == 0, , ],mapping = aes(tsne1, tsne2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene4>0, , ],mapping = aes(tsne1, tsne2, color=gene4)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[4], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
p8 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene4 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene4>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene4)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[4], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
sc_meta$gene5 <- as.numeric(seu_obj@assays$RNA@data[genes[5], ])
p9 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(tsne1, tsne2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene5 == 0, , ],mapping = aes(tsne1, tsne2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene5>0, , ],mapping = aes(tsne1, tsne2, color=gene5)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('t-SNE: ', genes[5], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
p10 <- ggplot(sc_meta) +
geom_point(sc_meta[sc_meta$scSpace != '11', ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene5 == 0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
geom_point(sc_meta[sc_meta$scSpace == '11'& sc_meta$gene5>0, , ],mapping = aes(pseudo_space1, pseudo_space2, color=gene5)) +
scale_color_gradient(low = 'grey90', high = 'blue') +
theme_classic() +
ggtitle(paste0('Pseudo space: ', genes[5], ' (C12)')) +
theme(plot.title = element_text(hjust = 0.5))
cowplot::plot_grid(p1, p3, p5, p7, p9, p2, p4, p6, p8, p10, ncol = 5)
Increasing the cluster resolution of Seurat
seu_obj$n14 <- seu_obj$seurat_clusters
# 15 clusters
seu_obj <- FindClusters(seu_obj, resolution = 0.445)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.9321
## Number of communities: 15
## Elapsed time: 0 seconds
seu_obj$n15 <- seu_obj$seurat_clusters
# 16 clusters
seu_obj <- FindClusters(seu_obj, resolution = 0.45)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.9318
## Number of communities: 16
## Elapsed time: 0 seconds
seu_obj$n16 <- seu_obj$seurat_clusters
# 17 clusters
seu_obj <- FindClusters(seu_obj, resolution = 0.6)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.9190
## Number of communities: 17
## Elapsed time: 0 seconds
seu_obj$n17 <- seu_obj$seurat_clusters
# 18 clusters
seu_obj <- FindClusters(seu_obj, resolution = 0.7)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.9108
## Number of communities: 18
## Elapsed time: 0 seconds
seu_obj$n18 <- seu_obj$seurat_clusters
# 19 clusters
seu_obj <- FindClusters(seu_obj, resolution = 0.9)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.8963
## Number of communities: 19
## Elapsed time: 0 seconds
seu_obj$n19 <- seu_obj$seurat_clusters
# 20 clusters
seu_obj <- FindClusters(seu_obj, resolution = 0.97)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.8915
## Number of communities: 20
## Elapsed time: 0 seconds
seu_obj$n20 <- seu_obj$seurat_clusters
# 21 clusters
seu_obj <- FindClusters(seu_obj, resolution = 1)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.8896
## Number of communities: 21
## Elapsed time: 0 seconds
seu_obj$n21 <- seu_obj$seurat_clusters
# 22 clusters
seu_obj <- FindClusters(seu_obj, resolution = 1.4)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.8637
## Number of communities: 22
## Elapsed time: 0 seconds
seu_obj$n22 <- seu_obj$seurat_clusters
# 23 clusters
seu_obj <- FindClusters(seu_obj, resolution = 1.5)
## Modularity Optimizer version 1.3.0 by Ludo Waltman and Nees Jan van Eck
##
## Number of nodes: 3717
## Number of edges: 114436
##
## Running Louvain algorithm...
## Maximum modularity in 10 random starts: 0.8583
## Number of communities: 23
## Elapsed time: 0 seconds
seu_obj$n23 <- seu_obj$seurat_clusters
sc_meta <- seu_obj@meta.data
sc_meta$tsne1 <- seu_obj@reductions$tsne@cell.embeddings[, 1]
sc_meta$tsne2 <- seu_obj@reductions$tsne@cell.embeddings[, 2]
card_meta <- sc_meta[sc_meta$tsne1 >= (-7 - 20) & sc_meta$tsne1 < (-7 + 20)
& sc_meta$tsne2 >= (33 - 20) & sc_meta$tsne2 < (33 + 20), ]
p <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = celltype)) +
geom_circle(aes(x0=2.5, y0=36, r=6.5), color="black", linetype=2) +
scale_color_manual(values = c('#EF4444', '#A5B4FC', '#10B981')) +
theme_classic() +
ggtitle(paste0('Ground truth')) +
theme(plot.title = element_text(hjust = 0.5))
p
n14_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n14), 3)
n15_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n15), 3)
n16_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n16), 3)
n17_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n17), 3)
n18_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n18), 3)
n19_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n19), 3)
n20_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n20), 3)
n21_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n21), 3)
n22_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n22), 3)
n23_ari <- round(adjustedRandIndex(card_meta$celltype, card_meta$n23), 3)
p1 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n14)) +
scale_color_manual(values = c('#10B981', '#A5B4FC')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 14 (ARI: ', n14_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p2 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n15)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 15 (ARI: ', n15_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p3 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n16)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 16 (ARI: ', n16_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p4 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n17)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444', '#854D0E')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 17 (ARI: ', n17_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p5 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n18)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444', '#854D0E')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 18 (ARI: ', n18_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p6 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n19)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444', '#854D0E')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 19 (ARI: ', n19_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p7 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n20)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444', '#854D0E')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 20 (ARI: ', n20_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p8 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n21)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444', '#854D0E')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 21 (ARI: ', n21_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p9 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n22)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444', '#854D0E')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 22 (ARI: ', n22_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
p10 <- ggplot() +
geom_point(card_meta, mapping = aes(tsne1, tsne2, color = n23)) +
scale_color_manual(values = c('#10B981', '#A5B4FC', '#EF4444', '#8B5CF6', '#854D0E')) +
theme_classic() +
theme(legend.position = 'none') +
ggtitle(paste0('N = 23 (ARI: ', n23_ari, ')')) +
theme(plot.title = element_text(hjust = 0.5))
cowplot::plot_grid(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10)
# the change of ARI as the k increase
ari_all <- data.frame(
value = c(
adjustedRandIndex(sc_meta$celltype, sc_meta$scSpace),
adjustedRandIndex(sc_meta$celltype, sc_meta$n14),
adjustedRandIndex(sc_meta$celltype, sc_meta$n15),
adjustedRandIndex(sc_meta$celltype, sc_meta$n16),
adjustedRandIndex(sc_meta$celltype, sc_meta$n17),
adjustedRandIndex(sc_meta$celltype, sc_meta$n18),
adjustedRandIndex(sc_meta$celltype, sc_meta$n19),
adjustedRandIndex(sc_meta$celltype, sc_meta$n20),
adjustedRandIndex(sc_meta$celltype, sc_meta$n21),
adjustedRandIndex(sc_meta$celltype, sc_meta$n22),
adjustedRandIndex(sc_meta$celltype, sc_meta$n23),
adjustedRandIndex(card_meta$celltype, card_meta$scSpace),
adjustedRandIndex(card_meta$celltype, card_meta$n14),
adjustedRandIndex(card_meta$celltype, card_meta$n15),
adjustedRandIndex(card_meta$celltype, card_meta$n16),
adjustedRandIndex(card_meta$celltype, card_meta$n17),
adjustedRandIndex(card_meta$celltype, card_meta$n18),
adjustedRandIndex(card_meta$celltype, card_meta$n19),
adjustedRandIndex(card_meta$celltype, card_meta$n20),
adjustedRandIndex(card_meta$celltype, card_meta$n21),
adjustedRandIndex(card_meta$celltype, card_meta$n22),
adjustedRandIndex(card_meta$celltype, card_meta$n23)),
method = c('scSpace', rep('Seurat', 10), 'scSpace', rep('Seurat', 10)),
ct_num = c(14, 14:23, 14, 14:23),
type = c(rep('Seurat (all)', 11), rep('Seurat (cardiomyocytes)', 11))
)
write.csv(df, file = paste0(output_dir, '/ARI_for_different_cluster_numbers.csv'))
ari_all$ct_num <- as.factor(ari_all$ct_num)
# ari_all[ari_all$method == 'scSpace' & ari_all$type == 'all celltypes', ]$value
# 0.7303123
# ari_all[ari_all$method == 'scSpace' & ari_all$type == 'three cardiomyocytes', ]$value
# 0.907676
p <- ggplot(ari_all[ari_all$method != 'scSpace', ],
aes(x = ct_num, y = value, color = type, group = type)) +
geom_point(size = 3) +
geom_line(size = 1) +
geom_hline(yintercept = 0.7303123, linetype=2, size = 1, color = '#F8766D') +
geom_hline(yintercept = 0.907676, linetype=2, size = 1, color = '#00BFC4') +
annotate(geom="text", x = 8, y = 0.77, label = 'scSpace (all)', colour='#F8766D') +
annotate(geom="text", x = 8, y = 0.95, label = 'scSpace (cardiomyocytes)', colour='#00BFC4') +
ylim(c(0, 1)) +
ylab('ARI') +
xlab('Cluste number') +
theme_classic()
## Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
## ℹ Please use `linewidth` instead.
p
