---
title: "Validation_FMN1_RBFOX1_OLIG2_IF_manual_quantification"
author: "Luise A. Seeker"
date: "09/08/2022"
output: html_document
---
This is based on only three stainings BA4 samples (I did quantify 1 CSC sample
but there is so much FMN1 in the background in the CSC that I think it is not
reliable). FMN1 staining
looks very interesting, particularly in the brain where it almost marks tiling
cells of unclear identity that I think may be microglia based in them mostly
being OLIG2 negative and based on the RNAseq data that also marks neurons and
microglia as expressing FMN1 but not astrocytes.
OPCs and COPs also do not express FMN1 which means it may be responsible
for cytoskeleton reshaping in other situations than myelination (where the
Arp2/3 complex is important). Maybe it is important for re-myelinaton?
So, what I need to do here is to only show that FMN+ olig2+ cells exist which
is more tricky than it sounds, because FMN1 is located in processes.
```{r}
library(ggplot2)
library(lme4)
library(lmerTest)
library(here)
library(tidyr)
library(ggsci)
library(Seurat)
```
```{r, echo = F}
mypal <- pal_npg("nrc", alpha = 0.7)(10)
mypal2 <-pal_tron("legacy", alpha = 0.7)(7)
mypal3 <- pal_lancet("lanonc", alpha = 0.7)(9)
mypal4 <- pal_simpsons(palette = c("springfield"), alpha = 0.7)(16)
mypal5 <- pal_rickandmorty(palette = c("schwifty"), alpha = 0.7)(6)
mypal6 <- pal_futurama(palette = c("planetexpress"), alpha = 0.7)(5)
mypal7 <- pal_startrek(palette = c("uniform"), alpha = 0.7)(5)
mycoloursP<- c(mypal, mypal2, mypal3, mypal4, mypal5, mypal6, mypal7)
```
```{r}
fmn1_rbfox1 <- read.csv(here("data",
"validation_data",
"20220818_FMN1_RBFOX1_OLIG2_manual.csv"))
```
```{r}
names(fmn1_rbfox1)
fmn1_rbfox1$tissue <- as.factor(fmn1_rbfox1$tissue)
fmn1_rbfox1$sample_id <- factor(fmn1_rbfox1$sample_id, levels = c("SD021_17_BA4",
"SD031_14_BA4",
"SD042_18_BA4",
"SD008_14_CSC"))
```
Convert from wide to long format to see how many single, double and triple
positive cells are in each sample
```{r}
keycol <- "sample_id"
valuecol <- "count"
gathercols <- c("FMN1_pos",
"OLIG2_pos",
"RBFOX1_pos",
"FMN1_OLIG2",
"FMN1_RBFOX1",
"RBFOX1_OLIG2",
"FMN1_RBFOX1_OLIG2",
"negative")
long_data <- gather(fmn1_rbfox1, keycol, valuecol, gathercols)
head(long_data)
nrow(long_data)
names(long_data)
```
```{r}
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
Remove cells that are negative for all tested markers:
```{r}
keycol <- "sample_id"
valuecol <- "count"
gathercols <- c("FMN1_pos",
"OLIG2_pos",
"RBFOX1_pos",
"FMN1_OLIG2",
"FMN1_RBFOX1",
"RBFOX1_OLIG2",
"FMN1_RBFOX1_OLIG2")
long_data <- gather(fmn1_rbfox1, keycol, valuecol, gathercols)
head(long_data)
nrow(long_data)
names(long_data)
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
Exclude cells that are not positive for OLIG2
```{r}
keycol <- "sample_id"
valuecol <- "count"
gathercols <- c("OLIG2_pos",
"FMN1_OLIG2",
"RBFOX1_OLIG2",
"FMN1_RBFOX1_OLIG2")
long_data <- gather(fmn1_rbfox1, keycol, valuecol, gathercols)
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
collapse tissue
```{r}
ggplot(long_data, aes(x = tissue ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(long_data, aes(x = tissue,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
collapse all samples BA4 and CSC
```{r}
long_data$all_samples <- "all_samples"
ggplot(long_data, aes(x = all_samples ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(long_data, aes(x = all_samples,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
remove OLIG2_pos that are negative for both RBFOX1 and FMN1
```{r}
marker_dat <- subset(long_data, long_data$keycol != "OLIG2_pos")
ggplot(marker_dat, aes(x = all_samples,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
BA4 only
```{r}
subs_data <- subset(long_data, long_data$tissue == "BA4")
ggplot(subs_data, aes(x = tissue ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(subs_data, aes(x = tissue,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
```{r}
subs_subs_data <- subset(subs_data, subs_data$keycol != "OLIG2_pos")
ggplot(subs_subs_data, aes(x = tissue,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
```{r}
summary(fmn1_rbfox1$FMN1_oligos_perc_total_oligos)
```
```{r}
nad_ol <- readRDS(here("data",
"single_nuc_data",
"oligodendroglia",
"srt_oligos_and_opcs_LS.RDS"))
Idents(nad_ol) <- "ol_clusters_named"
```
FMN1 and RBFOX1 co-expression in snRNAseq
```{r}
oligos <- subset(nad_ol, idents = c("Oligo_A",
"Oligo_B",
"Oligo_C",
"Oligo_D",
"Oligo_E",
"Oligo_F"))
Idents(oligos) <- "Tissue"
ba4_ol <-subset(oligos, idents = "BA4")
cb_ol <-subset(oligos, idents = "CB")
csc_ol <-subset(oligos, idents = "CSC")
FMN1 <- 'FMN1'
RBFOX1 <- 'RBFOX1'
FMN1.cutoff <- 1
RBFOX1.cutoff <- 1
#BA4
FMN1_cells_ba4 <- length(which(FetchData(ba4_ol, vars = FMN1) > FMN1.cutoff &
FetchData(ba4_ol, vars = RBFOX1) < RBFOX1.cutoff))
RBFOX1_cells_ba4 <- length(which(FetchData(ba4_ol, vars = RBFOX1) > RBFOX1.cutoff&
FetchData(ba4_ol, vars = FMN1) < FMN1.cutoff))
FMN1_RBFOX1_olig2_cells_ba4 <- length(which(FetchData(ba4_ol, vars = RBFOX1) > RBFOX1.cutoff &
FetchData(ba4_ol, vars = FMN1) > FMN1.cutoff))
all_cells_incluster_ba4 <- table(ba4_ol@active.ident)
FMN1_ba4 <-FMN1_cells_ba4/all_cells_incluster_ba4 * 100
# Percentage of cells in dataset that express FMN1
RBFOX1_ba4 <- RBFOX1_cells_ba4/all_cells_incluster_ba4 * 100
#Percentage of cells in dataset that express RBFOX1
double_ba4 <- FMN1_RBFOX1_olig2_cells_ba4/all_cells_incluster_ba4 * 100
#Percentage of cells in dataset that co-express FMN1 + RBFOX1
#CB
FMN1_cells_cb <- length(which(FetchData(cb_ol, vars = FMN1) > FMN1.cutoff&
FetchData(cb_ol, vars = RBFOX1) < RBFOX1.cutoff))
RBFOX1_cells_cb <- length(which(FetchData(cb_ol, vars = RBFOX1) > RBFOX1.cutoff &
FetchData(cb_ol, vars = FMN1) < FMN1.cutoff))
FMN1_RBFOX1_olig2_cells_cb <- length(which(FetchData(cb_ol, vars = RBFOX1) >
RBFOX1.cutoff & FetchData(cb_ol, vars = FMN1) >
FMN1.cutoff))
all_cells_incluster_cb <- table(cb_ol@active.ident)
FMN1_cb <- FMN1_cells_cb/all_cells_incluster_cb * 100 # Percentage of cells in dataset that express FMN1
RBFOX1_cb <- RBFOX1_cells_cb/all_cells_incluster_cb * 100 #Percentage of cells in dataset that express RBFOX1
double_cb <- FMN1_RBFOX1_olig2_cells_cb/all_cells_incluster_cb * 100 #Percentage of cells in dataset that co-express FMN1 + RBFOX1
#CSC
FMN1_cells_csc <- length(which(FetchData(csc_ol, vars = FMN1) > FMN1.cutoff&
FetchData(csc_ol, vars = RBFOX1) < RBFOX1.cutoff))
RBFOX1_cells_csc <- length(which(FetchData(csc_ol, vars = RBFOX1) > RBFOX1.cutoff &
FetchData(csc_ol, vars = FMN1) < FMN1.cutoff))
FMN1_RBFOX1_olig2_cells_csc <- length(which(FetchData(csc_ol, vars = RBFOX1) >
RBFOX1.cutoff & FetchData(csc_ol, vars = FMN1) >
FMN1.cutoff))
all_cells_incluster_csc <- table(csc_ol@active.ident)
FMN1_csc <- FMN1_cells_csc/all_cells_incluster_csc * 100
# Percentage of cells in dataset that express FMN1
RBFOX1_csc <- RBFOX1_cells_csc/all_cells_incluster_csc * 100
#Percentage of cells in dataset that express RBFOX1
double_csc <- FMN1_RBFOX1_olig2_cells_csc/all_cells_incluster_csc * 100
#Percentage of cells in dataset that co-express FMN1 + RBFOX1
df_snrna_seq <- data.frame(Tissue = c(rep("BA4", 4),
rep("CB", 4),
rep("CSC", 4)),
gene_expr = rep(c("FMN1+",
"RBFOX1+",
"FMN1+RBFOX1+",
"Other Oligodendrocytes"), 3),
percentage = c(as.numeric(FMN1_ba4),
as.numeric(RBFOX1_ba4),
as.numeric(double_ba4),
100 - (as.numeric(FMN1_ba4)+
as.numeric(RBFOX1_ba4)+
as.numeric(double_ba4)),
as.numeric(FMN1_cb),
as.numeric(RBFOX1_cb),
as.numeric(double_cb),
100 - (as.numeric(FMN1_cb)+
as.numeric(RBFOX1_cb)+
as.numeric(double_cb)),
as.numeric(FMN1_csc),
as.numeric(RBFOX1_csc),
as.numeric(double_csc),
100 - (as.numeric(FMN1_csc)+
as.numeric(RBFOX1_csc)+
as.numeric(double_csc))))
ggplot(df_snrna_seq, aes(x = Tissue, y=percentage, fill = gene_expr))+
geom_bar(stat="identity")+scale_fill_manual(values= mycoloursP[24:50])
```
```{r}
subs_df_snrna_seq <- subset(df_snrna_seq, df_snrna_seq$gene_expr !=
"Other Oligodendrocytes")
ggplot(subs_df_snrna_seq, aes(x = Tissue, y=percentage, fill = gene_expr))+
geom_bar(stat="identity")+scale_fill_manual(values= mycoloursP[25:50])
ggplot(subs_df_snrna_seq, aes(x = Tissue, y=percentage, fill = gene_expr))+
geom_bar(stat="identity", position = "fill")+scale_fill_manual(values= mycoloursP[25:50])
```
Simplify further to have only FMN1 positive oligos regardless of RBFOX1 expression
```{r}
keycol <- "sample_id"
valuecol <- "count"
gathercols <- c("OLIG2_pos",
"FMN1_OLIG2_sum")
long_data <- gather(fmn1_rbfox1, keycol, valuecol, gathercols)
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(long_data, aes(x = sample_id ,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
BA4 only
```{r}
subs_data <- subset(long_data, long_data$tissue == "BA4")
ggplot(subs_data, aes(x = tissue ,
y = valuecol,
fill = keycol)) +
geom_bar(position="stack", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1))+scale_fill_manual(values= mycoloursP[25:50])
ggplot(subs_data, aes(x = tissue,
y = valuecol,
fill = keycol)) +
geom_bar(position="fill", stat="identity")+ theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) +scale_fill_manual(values= mycoloursP[25:50])
```
Conclusion:
I found that automatically counting FMN1 positive cells lead to a large variation,
and hoped manual counting would alleviate this. But I still observe a lot of
variation among samples. The FMN1 validation is tricky, because it is
expressed in processes and it is sometimes difficult to decide if a process
belongs to a cell or merely touches it. I think what I can say is that a subset
of oligodendrocytes do express FMN1 and that differences in their percentage
between snRNAseq and IF may be due to technical differences, sample differences
or inaccuracy in quantifying IF images due to FMN1 being expressed in processes.
It was particularly difficult to quantify in Spinal cord due to staining of background structures (ECM/ cytoskelleton).
```{r}
sessionInfo()
```