I was using Seurat to analyse single cell RNA-seq data and I managed to draw a heatmap plot with DoHeatmap() after clustering and marker selection, but got a bunch of random characters appearing in the legend. They are random characters as they will change every time you run the code. I was worrying over it's something related to my own dataset, so I then tried the test Seurat object 'ifnb' but still got the same issue (see the red oval in the example plot).
example plot
I also tried importing the Seurat object in R in the terminal (via readRDS) and ran the plotting function, but got the same issue there, so it's not a Rstudio thing.
Here are the codes I ran:
'''
library(Seurat)
library(SeuratData)
library(patchwork)
InstallData("ifnb")
LoadData("ifnb")
ifnb.list <- SplitObject(ifnb, split.by = "stim")
ifnb.list <- lapply(X = ifnb.list, FUN = function(x) {
x <- NormalizeData(x)
x <- FindVariableFeatures(x, selection.method = "vst", nfeatures = 2000)
})
features <- SelectIntegrationFeatures(object.list = ifnb.list)
immune.anchors <- FindIntegrationAnchors(object.list = ifnb.list, anchor.features = features)
immune.combined <- IntegrateData(anchorset = immune.anchors)
immune.combined <- ScaleData(immune.combined, verbose = FALSE)
immune.combined <- RunPCA(immune.combined, npcs = 30, verbose = FALSE)
immune.combined <- RunUMAP(immune.combined, reduction = "pca", dims = 1:30)
immune.combined <- FindNeighbors(immune.combined, reduction = "pca", dims = 1:30)
immune.combined <- FindClusters(immune.combined, resolution = 0.5)
DefaultAssay(immune.combined) <- 'RNA'
immune_markers <- FindAllMarkers(immune.combined, latent.vars = "stim", test.use = "MAST", assay = 'RNA')
immune_markers %>%
group_by(cluster) %>%
top_n(n = 10, wt = avg_log2FC) -> top10_immune
DoHeatmap(immune.combined, slot = 'data',features = top10_immune$gene, group.by = 'stim', assay = 'RNA')
'''
Does anyone have any idea how to solve this issue other than reinstalling everything?
I have been having the same issue myself and while I have solved it by not needing the legend, I think you could use this approach and use a similar solution:
DoHeatmap(immune.combined, slot = 'data',features = top10_immune$gene, group.by = 'stim', assay = 'RNA') +
scale_color_manual(
values = my_colors,
limits = c('CTRL', 'STIM'))
Let me know if this works! It doesn't solve the source of the odd text values but it does the job! If you haven't already, I would recommend creating a forum question on the Seurat forums to see where these characters are coming from!
When I use seurat4.0, I met the same problem.
While I loaded 4.1, it disappeared
Related
I am new to Seurat, and am trying to run an integrated analysis of two different single-nuclei RNAseq datasets. I have been following the Seurat tutorial on integrated analysis (https://satijalab.org/seurat/articles/integration_introduction.html) to guide me, but when I ran the last line of code, I got an error.
# Loading required libraries
library(Seurat)
library(cowplot)
library(patchwork)
# Set up the Seurat Object
vgat.data <- Read10X(data.dir = "~/Desktop/VGAT Viral Data 1/")
vglut.data <- Read10X(data.dir = "~/Desktop/VGLUT3 Viral/")
# Initialize the Seurat object with the raw (non-normalized data)
vgat <- CreateSeuratObject(counts = vgat.data, project = "VGAT/VGLUT Integration", min.cells = 3, min.features = 200)
vglut <- CreateSeuratObject(counts = vglut.data, project = "VGAT/VGLUT Integration", min.cells = 3, min.features = 200)
# Merging the datasets
vgat <- AddMetaData(vgat, metadata = "VGAT", col.name = "Cell")
vglut <- AddMetaData(vglut, metadata = "VGLUT", col.name = "Cell")
merged <- merge(vgat, y = vglut, add.cell.ids = c("VGAT", "VGLUT"), project = "VGAT/VGLUT Integration")
# Split the dataset into a list of two seurat objects (vgat and vglut)
merged.list <- SplitObject(merged, split.by = "Cell")
# Normalize and Identify variable features for each dataset independently
merged.list <lapply(X = merged.list, FUN = function(x) {
x <- NormalizeData(x)
x <- FindVariableFeatures(x, selection.method = "vst", nFeatures = 2000)
})
After running the last line of code, I get the following error: Error in merged.list < lapply(X = merged.list, FUN = function(x) { :
comparison of these types is not implemented
I was wondering if anyone is familiar with Seurat and knows how I can troubleshoot this error. Any help would be greatly appreciated.
I have 11 plots and used a looping function to plot them see my code below. However, I can't get them to fit in just 1 page or less. The plots are actually too big. I am using R software and writing my work in RMarkdown. I have spent almost an entire week trying to resolve this.
group_by(Firm_category) %>%
doo(
~ggboxplot(
data =., x = "Means.type", y = "means",
fill ="grey", palette = "npg", legend = "none",
ggtheme = theme_pubr()
),
result = "plots"
)
graph3
# Add statistical tests to each corresponding plot
Firm_category <- graph3$Firm_category
xx <- for(i in 1:length(Firm_category)){
graph3.i <- graph3$plots[[i]] +
labs(title = Firm_category[i]) +
stat_pvalue_manual(stat.test[i, ], label = "p.adj.signif")
print(graph3.i)
}
#output3.long data sample below as comments
#Firm_category billmonth Means.type means
#Agric 1 Before 38.4444
#Agric 1 After 51.9
Complete data is on my github: https://github.com/Fridahnyakundi/Descriptives-in-R/blob/master/Output3.csv
This code prints all the graphs but in like 4 pages. I want to group them into a grid. I have tried to add all these codes below just before my last curly bracket and none is working, please help me out.
library(cowplot)
print(plot_grid(plotlist = graph3.i[1:11], nrow = 4, ncol = 3))
library(ggpubr)
print(ggarrange(graph3.i[1:11], nrow = 4, ncol = 3))
I tried the gridExtra command as well (they all seem to do the same thing). I am the one with a mistake and I guess it has to do with my list. I read a lot of similar work here, some suggested
dev.new()
dev.off()
I still didn't get what they do. But adding either of them caused my code to stop.
I tried defining my 'for' loop function say call it 'XX', then later call it to make a list of graph but it returned NULL output.
I have tried defining an empty list (as I read in some answers here) then counting them to make a list that can be printed but I got so many errors.
I have done this for almost 3 days and will appreciate your help in resolving this.
Thanks!
I tried to complete your code ... and this works (but I don't have your 'stat.test' object). Basically, I added a graph3.i <- list() and replaced graph3.i in the loop ..
Is it what you wanted to do ?
library(magrittr)
library(dplyr)
library(rstatix)
library(ggplot2)
library(ggpubr)
data <- read.csv(url('http://raw.githubusercontent.com/Fridahnyakundi/Descriptives-in-R/master/Output3.csv'))
graph3 <- data %>% group_by(Firm_category) %>%
doo(
~ggboxplot(
data =., x = "Means.type", y = "means",
fill ="grey", palette = "npg", legend = "none",
ggtheme = theme_pubr()
),
result = "plots"
)
graph3
# Add statistical tests to each corresponding plot
graph3.i <- list()
Firm_category <- graph3$Firm_category
xx <- for(i in 1:length(Firm_category)){
graph3.i[[i]] <- graph3$plots[[i]] +
labs(title = Firm_category[i]) # +
# stat_pvalue_manual(stat.test[i, ], label = "p.adj.signif")
print(graph3.i)
}
library(cowplot)
print(plot_grid(plotlist = graph3.i[1:11], nrow = 4, ncol = 3))
I want to report percentage by rows using CreateTableOne from tableone package in r using the following code
vars<-c( "Group1vsGroup2", "Single.Institution",
"Internal.Funding", "National.Funding",
"Industr1.Funding")
CatVar<-c( "Single.Institution",
"Internal.Funding", "National.Funding",
"Industr1.Funding")
tab2 <- CreateTableOne(vars = vars, strata = "Group1vsGroup2",factorVars = CatVar,data = df,test = T);tab2<-print(tab2, margin=1,test = T, varLabels = T,quote = T,dropEqual = T)
I added "margin=1" as I saw in this website although it was for tableone command (not CreateTableOne), but I got % by columns as in the image below. Any advice will be greatly appreciated.
Here a MWE you can start with, see ?Gmisc::getDescriptionStatsBy for more details.
library(Gmisc)
# Define cyl as a factor so that getDescriptionStatsBy can use the correct Gmisc::describe*
mtcars$cyl=as.factor(mtcars$cyl)
getTable1Stats <- function(x, digits = 0,...){
getDescriptionStatsBy(x = x,
by = mtcars$am,
digits = digits,
header_count = TRUE,
...)
}
t1 <- list()
t1[["Gas"]] <- getTable1Stats(mtcars$mpg, add_total_col="last")
#hrzl_prop=FALSE will indicate that the proportions are to be interpreted in a vertical manner
t1[["Cylinder†"]] <- getTable1Stats(mtcars$cyl, hrzl_prop=TRUE, add_total_col="last")
t1[["Disp"]] <- getTable1Stats(mtcars$disp, continuous_fn=describeMedian, add_total_col="last")
mergeDesc(t1,
htmlTable_args = list(css.rgroup = "",
caption = "Basic descriptive statistics from the mtcars dataset",
tfoot = "†The weight is in 10<sup>3</sup> kg")
)
PS: This solution based on Max's vignette here. For more details on htmlTable you can see its vignette here
I'm trying to optimize the parameters for baseline in the R baseline package by changing each parameters in a loop and comparing plots to determine which parameters give me the best baseline.
I currently have the code written so that the loop produces each plot, but I'm having trouble with getting the plot saved as the class of each object I'm creating is a baseline package-specific (which I'm suspecting is the problem here).
foo <- data.frame(Date=seq.Date(as.Date("1957-01-01"), by = "day",
length.out = ncol(milk$spectra)),
Visits=milk$spectra[1,],
Old_baseline_visits=milk$spectra[1,], row.names = NULL)
foo.t <- t(foo$Visits)
#the lines above were copied from https://stackoverflow.com/questions/37346967/r-packagebaseline-application-to-sample-dataset to make a reproducible dataset
df <- expand.grid(lambda=seq(1,10,1), p=seq(0.01,0.1,0.01))
baselinediff <- list()
for(i in 1:nrow(df)){
thislambda <- df[i,]$lambda
thisp <- df[i,]$p
thisplot <- baseline(foo.t, lambda=thislambda, p=thisp, maxit=20, method='als')
print(paste0("lambda = ", thislambda))
print(paste0("p = ", thisp))
print(paste0("index = ", i))
baselinediff[[i]] <- plot(thisplot)
jpeg(file = paste(baselinediff[[i]], '.jpeg', sep = ''))
dev.off()
}
I know that I would be able to extract corrected spectra using baseline.als but I just want to save the plot images with the red baseline so that I can see how well the baselines are getting drawn. Any baseline users out there that can help?
I suggest you change your loop in the following way:
for(i in 1:nrow(df)){
thislambda <- df[i,]$lambda
thisp <- df[i,]$p
thisplot <- baseline(foo.t, lambda=thislambda, p=thisp, maxit=20, method='als')
print(paste0("lambda = ", thislambda))
print(paste0("p = ", thisp))
print(paste0("index = ", i))
baselinediff[[i]] <- thisplot
jpeg(file = paste('baseline', i, '.jpeg', sep = ''))
plot(baselinediff[[i]])
dev.off()
}
Note that this does not try to capture the already plotted element (thisplot) inside of the list. Instead, the plotting is done after you call the jpeg command. This solves your export issue. Another problem was the naming of the file. If you call baselinediff[[i]] inside of paste, you apparently end up with an error. So I switched it to a simpler name. To plot your resulting list, call:
lapply(baselinediff, plot)
If you are determined on storing the already plotted element, the capture.plotfunction from the imager package might be a good start.
I'm trying to make very simple GUI for my script. In nutshell problem looks like that :
dataset is dataframe, I would like to plot one column as the time and use simple GUI for choosing next/previus column.
dataset <-data.frame(rnorm(10), rnorm(10), rnorm(10))
columnPlot <- function(dataset, i){
plot(dataset[, i])
}
how to use tcltk for calling fplot with different i's ?
Not what you asked for (not tcltkrelated), but I would advise you to have a look at the new shiny package from RStudio.
Are you particularly attached to the idea of using tcltk? I've been working on something similar using the gWidgets package and have had some success. According to it's CRAN site, "gWidgets provides a toolkit-independent API for building interactive GUIs". This package uses tcltk or GTK2 and I've been using the GTK2 portion. Here's a quick example of a GUI with a spinbutton for changing i. I also added a little fanciness to your function because you mentioned you would be plotting time series, so I made the x axis Time.
data<-data.frame(rnorm(11),rnorm(11),rnorm(11))
i = 1
fplot <- function(i, data = data){
library(ggplot2)
TimeStart <- as.Date('1/1/2012', format = '%m/%d/%Y')
plotdat <- data.frame(Value = data[ ,i], Time = seq(TimeStart,TimeStart + nrow(data) - 1, by = 1))
myplot <- ggplot(plotdat, aes(x = Time, y = Value))+
geom_line()
print(myplot)
}
library(gWidgets)
options(guiToolkit = 'RGtk2')
window <- gwindow ("Time Series Plots", visible = T)
notebook <- gnotebook (cont = window)
group1 <- ggroup(cont = notebook, label = "Choose i", horizontal=F)
ichooser <- gspinbutton(cont = group1, from = 1, to = ncol(data), by = 1, value = i, handler = function(h,...){
i <<- svalue(h$obj)})
plotbutton <- gbutton('Plot', cont = group1, handler=function(h,...){
fplot(i, data)})
graphicspane1 <- ggraphics(cont = group1)