Last updated: 2018-12-05

workflowr checks: (Click a bullet for more information)
  • R Markdown file: up-to-date

    Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.

  • Environment: empty

    Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

  • Seed: set.seed(20180730)

    The command set.seed(20180730) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

  • Session information: recorded

    Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

  • Repository version: bd6173a

    Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.

    Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:
    
    Ignored files:
        Ignored:    .DS_Store
        Ignored:    .Rhistory
        Ignored:    .Rproj.user/
        Ignored:    analysis/cache.bak.20181031/
        Ignored:    analysis/cache.bak/
        Ignored:    analysis/cache.lind2.20181114/
        Ignored:    analysis/cache/
        Ignored:    data/Lindstrom2/
        Ignored:    data/processed.bak.20181031/
        Ignored:    data/processed.bak/
        Ignored:    data/processed.lind2.20181114/
        Ignored:    output/04B_Organoids_Nephron/cluster_de.csv.zip
        Ignored:    output/04B_Organoids_Nephron/conserved_markers.csv.zip
        Ignored:    output/04B_Organoids_Nephron/markers.csv.zip
        Ignored:    output/04_Organoids_Clustering/cluster_de.csv.zip
        Ignored:    output/04_Organoids_Clustering/conserved_markers.csv.zip
        Ignored:    output/04_Organoids_Clustering/markers.csv.zip
        Ignored:    output/07B_Combined_Nephron/cluster_de.csv.zip
        Ignored:    output/07B_Combined_Nephron/cluster_de_filtered.csv.zip
        Ignored:    output/07B_Combined_Nephron/conserved_markers.csv.zip
        Ignored:    output/07B_Combined_Nephron/markers.csv.zip
        Ignored:    output/07B_Combined_Nephron/podocyte_de.csv.zip
        Ignored:    output/07B_Combined_Nephron/podocyte_de_filtered.csv.zip
        Ignored:    output/07_Combined_Clustering/cluster_de.csv.zip
        Ignored:    output/07_Combined_Clustering/cluster_de_filtered.csv.zip
        Ignored:    output/07_Combined_Clustering/conserved_markers.csv.zip
        Ignored:    output/07_Combined_Clustering/group_de.csv.zip
        Ignored:    output/07_Combined_Clustering/markers.csv.zip
        Ignored:    packrat/lib-R/
        Ignored:    packrat/lib-ext/
        Ignored:    packrat/lib/
        Ignored:    packrat/src/
        Ignored:    test.csv.zip
    
    Unstaged changes:
        Deleted:    output/04B_Organoids_Nephron/cluster_de.csv
        Modified:   output/04B_Organoids_Nephron/cluster_de.xlsx
        Deleted:    output/04B_Organoids_Nephron/conserved_markers.csv
        Modified:   output/04B_Organoids_Nephron/conserved_markers.xlsx
        Deleted:    output/04B_Organoids_Nephron/markers.csv
        Modified:   output/04B_Organoids_Nephron/markers.xlsx
        Deleted:    output/04_Organoids_Clustering/cluster_de.csv
        Modified:   output/04_Organoids_Clustering/cluster_de.xlsx
        Deleted:    output/04_Organoids_Clustering/conserved_markers.csv
        Modified:   output/04_Organoids_Clustering/conserved_markers.xlsx
        Deleted:    output/04_Organoids_Clustering/markers.csv
        Modified:   output/04_Organoids_Clustering/markers.xlsx
        Deleted:    output/07B_Combined_Nephron/cluster_de.csv
        Modified:   output/07B_Combined_Nephron/cluster_de.xlsx
        Deleted:    output/07B_Combined_Nephron/cluster_de_filtered.csv
        Modified:   output/07B_Combined_Nephron/cluster_de_filtered.xlsx
        Deleted:    output/07B_Combined_Nephron/conserved_markers.csv
        Modified:   output/07B_Combined_Nephron/conserved_markers.xlsx
        Deleted:    output/07B_Combined_Nephron/markers.csv
        Modified:   output/07B_Combined_Nephron/markers.xlsx
        Deleted:    output/07B_Combined_Nephron/podocyte_de.csv
        Deleted:    output/07B_Combined_Nephron/podocyte_de_filtered.csv
        Deleted:    output/07_Combined_Clustering/cluster_de.csv
        Modified:   output/07_Combined_Clustering/cluster_de.xlsx
        Deleted:    output/07_Combined_Clustering/cluster_de_filtered.csv
        Modified:   output/07_Combined_Clustering/cluster_de_filtered.xlsx
        Deleted:    output/07_Combined_Clustering/conserved_markers.csv
        Modified:   output/07_Combined_Clustering/conserved_markers.xlsx
        Deleted:    output/07_Combined_Clustering/group_de.csv
        Deleted:    output/07_Combined_Clustering/markers.csv
        Modified:   output/07_Combined_Clustering/markers.xlsx
    
    
    Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
Expand here to see past versions:
    File Version Author Date Message
    Rmd bd6173a Luke Zappia 2018-12-05 Add gene ids to output files
    html 582acea Luke Zappia 2018-12-03 Fix DE results summary plot cluster labels
    html a61f9c9 Luke Zappia 2018-09-13 Rebuild site
    html ad10b21 Luke Zappia 2018-09-13 Switch to GitHub
    Rmd 7755ac7 Luke Zappia 2018-08-15 Add methods document
    Rmd bff4d5b Luke Zappia 2018-08-14 Add crossover document
    Rmd 30718d3 Luke Zappia 2018-08-14 Add nephron reclustering


# scRNA-seq
library("Seurat")

# Plotting
library("clustree")
library("viridis")

# Presentation
library("glue")
library("knitr")

# Parallel
library("BiocParallel")

# Paths
library("here")

# Output
library("writexl")
library("jsonlite")

# Tidyverse
library("tidyverse")
source(here("R/output.R"))
orgs.path <- here("data/processed/Organoids_clustered.Rds")
bpparam <- MulticoreParam(workers = 10)

Introduction

In this document we are going to recluster the nephron clusters identified in the organoids analysis.

if (file.exists(orgs.path)) {
    orgs <- read_rds(orgs.path)
} else {
    stop("Clustered Organoids dataset is missing. ",
         "Please run '04_Organoids_Clustering.Rmd' first.",
         call. = FALSE)
}

Subsetting

clusters <- c(2, 9)
orgs.neph <- SubsetData(orgs, ident.use = clusters)
orgs.neph <- RunTSNE(orgs.neph, reduction.use = "cca.aligned", dims.use = 1:25)

We are going to select only the cells in clusters 2 and 9. This leaves us with 1125 cells.

Clustering

Selecting resolution

# Clear old clustering
not.res <- !grepl("res\\.", colnames(orgs.neph@meta.data))
orgs.neph@meta.data <- orgs.neph@meta.data[, not.res]

n.dims <- 25
resolutions <- seq(0, 1, 0.1)
orgs.neph <- FindClusters(orgs.neph, reduction.type = "cca.aligned",
                          dims.use = 1:n.dims, resolution = resolutions,
                          force.recalc = TRUE)

Seurat has a resolution parameter that indirectly controls the number of clusters it produces. We tried clustering at a range of resolutions from 0 to 1.

t-SNE plots

Here are t-SNE plots of the different clusterings.

src_list <- lapply(resolutions, function(res) {
    src <- c("#### Res {{res}} {.unnumbered}",
             "```{r cluster-tSNE-{{res}}}",
             "TSNEPlot(orgs.neph, group.by = 'res.{{res}}', do.return = TRUE)",  
             "```",
             "")
    knit_expand(text = src)
})

out <- knit_child(text = unlist(src_list), options = list(cache = FALSE))

Res 0

TSNEPlot(orgs.neph, group.by = 'res.0', do.return = TRUE)

Res 0.1

TSNEPlot(orgs.neph, group.by = 'res.0.1', do.return = TRUE)

Res 0.2

TSNEPlot(orgs.neph, group.by = 'res.0.2', do.return = TRUE)

Res 0.3

TSNEPlot(orgs.neph, group.by = 'res.0.3', do.return = TRUE)

Res 0.4

TSNEPlot(orgs.neph, group.by = 'res.0.4', do.return = TRUE)

Res 0.5

TSNEPlot(orgs.neph, group.by = 'res.0.5', do.return = TRUE)

Res 0.6

TSNEPlot(orgs.neph, group.by = 'res.0.6', do.return = TRUE)

Res 0.7

TSNEPlot(orgs.neph, group.by = 'res.0.7', do.return = TRUE)

Res 0.8

TSNEPlot(orgs.neph, group.by = 'res.0.8', do.return = TRUE)

Res 0.9

TSNEPlot(orgs.neph, group.by = 'res.0.9', do.return = TRUE)

Res 1

TSNEPlot(orgs.neph, group.by = 'res.1', do.return = TRUE)

Clustering tree

Standard

Coloured by clustering resolution.

clustree(orgs.neph)

Stability

Coloured by the SC3 stability metric.

clustree(orgs.neph, node_colour = "sc3_stability")

Gene expression

Coloured by the expression of some well-known kidney marker genes.

genes <- c("PECAM1", "CDH5", "MEIS1", "PDGFRA", "HMGB2", "CENPA", "SIX1",
           "DAPL1", "NPHS1", "PODXL", "S100A8", "TYROBP", "MAL", "EMX2",
           "LRP2", "GATA3", "SLC12A1", "SPINT2", "TUBB2B", "STMN2", "TTYH1",
           "HBA1", "HBG1")

is_present <- genes %in% rownames(orgs.neph@data)

The following genes aren’t present in this dataset and will be skipped: HBG1

src_list <- lapply(genes[is_present], function(gene) {
    src <- c("##### {{gene}} {.unnumbered}",
             "```{r clustree-{{gene}}}",
             "clustree(orgs.neph, node_colour = '{{gene}}',",
                      "node_colour_aggr = 'mean',",
             "exprs = 'scale.data') +",
             "scale_colour_viridis_c(option = 'plasma', begin = 0.3)",
             "```",
             "")
    knit_expand(text = src)
})

out <- knit_child(text = unlist(src_list), options = list(cache = FALSE))
PECAM1
clustree(orgs.neph, node_colour = 'PECAM1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-PECAM1-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

CDH5
clustree(orgs.neph, node_colour = 'CDH5',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-CDH5-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

MEIS1
clustree(orgs.neph, node_colour = 'MEIS1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-MEIS1-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

PDGFRA
clustree(orgs.neph, node_colour = 'PDGFRA',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-PDGFRA-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

HMGB2
clustree(orgs.neph, node_colour = 'HMGB2',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-HMGB2-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

CENPA
clustree(orgs.neph, node_colour = 'CENPA',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-CENPA-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

SIX1
clustree(orgs.neph, node_colour = 'SIX1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-SIX1-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

DAPL1
clustree(orgs.neph, node_colour = 'DAPL1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-DAPL1-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

NPHS1
clustree(orgs.neph, node_colour = 'NPHS1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-NPHS1-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

PODXL
clustree(orgs.neph, node_colour = 'PODXL',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-PODXL-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

S100A8
clustree(orgs.neph, node_colour = 'S100A8',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-S100A8-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

TYROBP
clustree(orgs.neph, node_colour = 'TYROBP',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-TYROBP-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

MAL
clustree(orgs.neph, node_colour = 'MAL',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-MAL-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

EMX2
clustree(orgs.neph, node_colour = 'EMX2',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-EMX2-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

LRP2
clustree(orgs.neph, node_colour = 'LRP2',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-LRP2-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

GATA3
clustree(orgs.neph, node_colour = 'GATA3',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-GATA3-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

SLC12A1
clustree(orgs.neph, node_colour = 'SLC12A1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-SLC12A1-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

SPINT2
clustree(orgs.neph, node_colour = 'SPINT2',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-SPINT2-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

TUBB2B
clustree(orgs.neph, node_colour = 'TUBB2B',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-TUBB2B-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

STMN2
clustree(orgs.neph, node_colour = 'STMN2',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-STMN2-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

TTYH1
clustree(orgs.neph, node_colour = 'TTYH1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

HBA1
clustree(orgs.neph, node_colour = 'HBA1',
node_colour_aggr = 'mean',
exprs = 'scale.data') +
scale_colour_viridis_c(option = 'plasma', begin = 0.3)

Expand here to see past versions of clustree-HBA1-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

Selected resolution

res <- 0.5
orgs.neph <- SetIdent(orgs.neph,
                      ident.use = orgs.neph@meta.data[, paste0("res.", res)])
n.clusts <- length(unique(orgs.neph@ident))

Based on these plots we will use a resolution of 0.5.

Clusters

Let’s have a look at the clusters on a t-SNE plot.

p1 <- TSNEPlot(orgs.neph, do.return = TRUE, pt.size = 0.5,
               group.by = "DatasetSample")
p2 <- TSNEPlot(orgs.neph, do.label = TRUE, do.return = TRUE, pt.size = 0.5)
plot_grid(p1, p2)

We can also look at the number of cells in each cluster.

plot.data <- orgs.neph@meta.data %>%
    select(Dataset, cluster = paste0("res.", res)) %>%
    mutate(cluster = factor(as.numeric(cluster))) %>%
    group_by(cluster, Dataset) %>%
    summarise(count = n()) %>%
    mutate(clust_total = sum(count)) %>%
    mutate(clust_prop = count / clust_total) %>%
    group_by(Dataset) %>%
    mutate(dataset_total = sum(count)) %>%
    ungroup() %>%
    mutate(dataset_prop = count / dataset_total)

ggplot(plot.data, aes(x = cluster, y = count, fill = Dataset)) +
    geom_col()

We are also interested in what proportions of the cells in each cluster come from each datasets (i.e. are there dataset specific clusters?).

ggplot(plot.data, aes(x = cluster, y = clust_prop, fill = Dataset)) +
    geom_col()

Alternatively we can look at what proportion of the cells in each dataset are in each cluster. If each dataset has the same distribution of cell types the heights of the bars should be the same.

ggplot(plot.data, aes(x = cluster, y = dataset_prop, fill = Dataset)) +
    geom_col(position = position_dodge(0.9))

Marker genes

Clustering is not very useful if we don’t know what cell types the clusters represent. One way to work that out is to look at marker genes, genes that are differentially expressed in one cluster compared to all other cells. Here we use the Wilcoxon rank sum test genes that are present in at least 10 percent of cells in at least one group (a cluster or all other cells).

markers <- bplapply(seq_len(n.clusts) - 1, function(cl) {
    cl.markers <- FindMarkers(orgs.neph, cl, logfc.threshold = 0, min.pct = 0.1,
                              print.bar = FALSE)
    cl.markers$cluster <- cl
    cl.markers$gene <- rownames(cl.markers)
    return(cl.markers)
}, BPPARAM = bpparam)

markers <- bind_rows(markers) %>%
    select(gene, cluster, everything())

Here we print out the top two markers for each cluster.

markers %>% group_by(cluster) %>% top_n(2, abs(avg_logFC)) %>% data.frame

A heatmap can give us a better view. We show the top five positive marker genes for each cluster.

top <- markers %>% group_by(cluster) %>% top_n(5, avg_logFC)

cols <- viridis(100)[c(1, 50, 100)]

DoHeatmap(orgs.neph, genes.use = top$gene, slim.col.label = TRUE,
          remove.key = TRUE, col.low = cols[1], col.mid = cols[2],
          col.high = cols[3])

By cluster

markers.list <- lapply(0:(n.clusts - 1), function(x) {
    markers %>%
        filter(cluster == x, p_val < 0.05) %>%
        dplyr::arrange(-avg_logFC) %>%
        select(Gene = gene, LogFC = avg_logFC, pVal = p_val)
})

names(markers.list) <- paste("Cluster", 0:(n.clusts - 1))
marker.summary <- markers.list %>%
    map2_df(names(markers.list), ~ mutate(.x, Cluster = .y)) %>%
    mutate(IsUp = LogFC > 0) %>%
    group_by(Cluster) %>%
    summarise(Up = sum(IsUp), Down = sum(!IsUp)) %>%
    mutate(Down = -Down) %>%
    gather(key = "Direction", value = "Count", -Cluster) %>%
    mutate(Cluster = factor(Cluster, levels = names(markers.list)))

ggplot(marker.summary,
       aes(x = fct_rev(Cluster), y = Count, fill = Direction)) +
    geom_col() +
    geom_text(aes(y = Count + sign(Count) * max(abs(Count)) * 0.07,
                  label = abs(Count)),
              size = 6, colour = "grey25") +
    coord_flip() +
    scale_fill_manual(values = c("#377eb8", "#e41a1c")) +
    ggtitle("Number of identified genes") +
    theme(axis.title = element_blank(),
          axis.line = element_blank(),
          axis.ticks = element_blank(),
          axis.text.x = element_blank(),
          legend.position = "bottom")

Expand here to see past versions of marker-cluster-counts-1.png:
Version Author Date
ad10b21 Luke Zappia 2018-09-13

We can also look at the full table of significant marker genes for each cluster.

src_list <- lapply(0:(n.clusts - 1), function(i) {
    src <- c("### {{i}} {.unnumbered}",
             "```{r marker-cluster-{{i}}}",
             "markers.list[[{{i}} + 1]]",
             "```",
             "")
    knit_expand(text = src)
})

out <- knit_child(text = unlist(src_list),
                  options = list(echo = FALSE, cache = FALSE))

0

1

2

3