analysis-manuscript.Rmd

---
title: "Extended data"
output:
  word_document:
    toc: yes
    toc_depth: '2'
  pdf_document:
    toc: yes
    toc_depth: 2
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(tidyverse)
library(here)
library(dabestr)
library(FSA)
library(rstatix)
library(see)
library(ggridges)
library(spatstat)
library(ggpubr)
library(rstatix)
library(knitr)
```

\newpage
# PD-vulnerable neurons are less resilient to cell stress induced by hydrogen peroxide

## Figure 2-1

Nuclei (DAPI) count

```{r echo=FALSE, fig.align="center", message=FALSE, warning=FALSE, paged.print=FALSE}

data <- 
  read_csv(here("derived-data", "count.csv")) %>% 
   mutate(hydrogen_peroxide = as.factor(hydrogen_peroxide)) %>% 
   mutate(neuron = as.factor(neuron)) %>% 
  mutate(neuron = fct_relevel(neuron, c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"))) %>% 
   mutate(type = case_when(
     neuron == "SN" ~ "pd-vulnerable", 
     neuron == "LC" ~ "pd-vulnerable",
     neuron == "R" ~ "pd-vulnerable",
     neuron == "DMV" ~ "pd-vulnerable", 
     neuron == "VTA" ~ "non-vulnerable",
     neuron == "XII" ~ "non-vulnerable", 
     neuron == "STR" ~ "non-vulnerable")) %>% 
  group_by(image, neuron) %>% 
  mutate(neuron_normalized = neuron_n / (mean(neuron_n[hydrogen_peroxide == 0], na.rm = TRUE))) %>% 
  mutate(dapi_normalized = dapi_n / (mean(dapi_n[hydrogen_peroxide == 0], na.rm = TRUE))) %>% 
  ungroup() %>% 
  select(neuron, hydrogen_peroxide, type, neuron_normalized, dapi_normalized)

```

```{r echo=FALSE, fig.align="center", fig.align="center", message=FALSE, warning=FALSE, paged.print=FALSE, out.width="50%"}

ggplot(data, aes(hydrogen_peroxide, dapi_normalized)) + 
  geom_point(aes(color = type), 
             width = 0.3, 
             size = 2, 
             alpha = 0.1, 
             position = position_dodge(width=1.3)) +
  theme_bw() +
  labs(
    x = "", 
    y = "normalized number of nuclei") +
  geom_boxplot(aes(color = type), 
               width = 0.2, 
               alpha = 0.01,
               position=position_dodge(width=0.8)) +
  theme(
    legend.position = "none") +
  theme_classic() + 
  geom_hline(
    yintercept=1, 
    color = "red", 
    size=0.5, 
    alpha = 0.2) +
  geom_boxplot(fill = "Black", width = 0.1, 
               alpha = 0.05, 
               outlier.alpha = 0) +
  ylim(0, 2)

ggsave(here("plots", "count_dapi.pdf"))

```

```{r eval=FALSE, error=FALSE, message=FALSE, warning=FALSE, comment=NA, include=FALSE, out.width="50%"}
#nomarlity test 

ggqqplot(data$dapi_normalized)

data %>%
  group_by(hydrogen_peroxide) %>% 
  shapiro_test(dapi_normalized)
```

Kurskal-Wallis

```{r, echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(dapi_normalized ~ hydrogen_peroxide, data = data)
```

Dunn test

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
out <- dunnTest(dapi_normalized ~ hydrogen_peroxide, data = data, method = "bonferroni")

out$res %>% 
  knitr::kable(align = "r") 
```

\newpage
Neuron count

## Figure 2-2

```{r echo=FALSE, fig.align="center", error = FALSE, message = FALSE, warning= FALSE, comment = NA, out.width="50%"}
ggplot(data, aes(hydrogen_peroxide, neuron_normalized)) + 
  geom_point(aes(color = type), 
             width = 0.3, 
             size = 2, 
             alpha = 0.1, 
             position = position_dodge(width=1.3)) +
  theme_bw() +
  labs(
    x = "", 
    y = "normalized number of 'positive' neurons") +
  geom_boxplot(aes(color = type), 
               width = 0.2, 
               alpha = 0.01,
               position=position_dodge(width=0.8)) +
  theme(
    legend.position = "none") +
  theme_classic() + 
  geom_hline(
    yintercept=1, 
    color = "red", 
    size=0.5, 
    alpha = 0.2) +
  geom_boxplot(fill = "Black", width = 0.1, 
               alpha = 0.1, 
               outlier.alpha = 0)

ggsave(here("plots", "count_neuron.pdf"))
```

Kurskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(neuron_normalized ~ hydrogen_peroxide, data = data)
```

Dunn test

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}

dunnTest(neuron_normalized ~ hydrogen_peroxide, data = data,
         method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

\newpage
# Differential vulnerability of neurons to hydrogen peroxide

## Figure 3-1

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data, aes(neuron, neuron_normalized, fill = neuron)) + 
  geom_jitter(width = 0.01, height = 0.3, size = 2, alpha = 0.1) +
  theme_bw() +
  labs(x = "", y = "normalized number of neurons") +
  geom_boxplot(aes(color = neuron), show.legend = FALSE, 
               width = 0.15, 
               alpha = 0.4, 
               position = position_nudge(x = -0.2),
               outlier.alpha = 0) +
  theme_classic() + 
  facet_wrap(~hydrogen_peroxide) +
  geom_hline(yintercept=1, 
                color = "red", size=0.5, alpha = 0.2) +
  scale_y_continuous(limits = c(0, 2.5), oob = scales::squish) +
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
 
        strip.background = element_rect(colour="NA", fill="#899DA4"))

ggsave(here("Plots", "count_neuron_compared.pdf"))
```


Kruskal-Wallis and Dunn test for each concetration

0 micromollar

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
#0 
kruskal.test(neuron_normalized ~ neuron, 
             data = data %>% 
               filter(hydrogen_peroxide == "0"))
  
dunnTest(neuron_normalized ~ neuron, 
         data = data %>% 
               filter(hydrogen_peroxide == "0"),
         method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

100 micromollar

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
# 100 
kruskal.test(neuron_normalized ~ neuron, 
             data = data %>% 
               filter(hydrogen_peroxide == "100"))
  
dunnTest(neuron_normalized ~ neuron, 
         data = data %>% 
               filter(hydrogen_peroxide == "100"),
         method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

150 micromollar

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
#150 
kruskal.test(neuron_normalized ~ neuron, 
             data = data %>% 
               filter(hydrogen_peroxide == "150"))
  
dunnTest(neuron_normalized ~ neuron, 
         data = data %>% 
               filter(hydrogen_peroxide == "150"),
         method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

200 micromollar

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
# 200 
kruskal.test(neuron_normalized ~ neuron, 
             data = data %>% 
               filter(hydrogen_peroxide == "200"))
  
dunnTest(neuron_normalized ~ neuron, 
         data = data %>% 
               filter(hydrogen_peroxide == "200"),
         method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

Estimation statistics

0 micromollar

```{r, echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data %>% 
  filter(hydrogen_peroxide == "0") %>% 
  drop_na() %>% 
  dabest(neuron, neuron_normalized, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot", 
     effsize.ylim = c(-0.5, 1))

ggsave(here("plots", "count_neuron_estimation_0.pdf"))

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c") 
```

100 micromollar

```{r, echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data %>% 
  filter(hydrogen_peroxide == "100") %>% 
  drop_na() %>% 
  dabest(neuron, neuron_normalized, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot", 
     effsize.ylim = c(-0.5, 1))

ggsave(here("plots", "count_neuron_estimation_100.pdf"))

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c") 
```

150 micromollar

```{r, echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data %>% 
  filter(hydrogen_peroxide == "150") %>% 
  drop_na() %>% 
  dabest(neuron, neuron_normalized, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot", 
     effsize.ylim = c(-0.5, 1))

ggsave(here("plots", "count_neuron_estimation_150.pdf"))

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c") 
```

200 micromollar

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment=NA, out.width="50%"}
shared.control <- 
  data %>% 
  filter(hydrogen_peroxide == "200") %>% 
  drop_na() %>% 
  dabest(neuron, neuron_normalized, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot", 
     effsize.ylim = c(-0.5, 1))

ggsave(here("plots", "count_neuron_estimation_200.pdf"))

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c") 
```

\newpage
# No overt difference in mitochondrial ROS production observed between neurons

## Figure 4-2

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}

data <- 
  read_csv(here("derived-data", "derived-roGFP.csv"))

```

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
data_for_plotting_2 <- 
data %>% 
  group_by(ROI, neuron, field, type) %>% 
  summarise(relative_oxidation = mean(relative_redox_state))

#https://easystats.github.io/see/reference/geom_violinhalf.html   

ggplot(data_for_plotting_2, aes(neuron, relative_oxidation)) +
  theme_classic() +
  geom_boxplot(width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(data = . %>% filter(field == "axonal"), color = NA, alpha = 0.3, aes(fill = field)) +
  geom_violinhalf(data = . %>% filter(field == "std"), color = NA, alpha = 0.3, aes(fill = field))

ggsave(here("plots", "roGFP_1.pdf"))

ggplot(data_for_plotting_2, aes(neuron, relative_oxidation)) +
  theme_classic() +
  geom_violinhalf(width = .5, fill = "black", alpha = 0.1) 

ggsave(here("plots", "roGFP_4.pdf"))
```

Kruskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(relative_oxidation ~ neuron, 
             data = data_for_plotting_2)
```

Dunn

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
dunnTest(relative_oxidation ~ neuron, 
             data = data_for_plotting_2, method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

Estimation statistics

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data_for_plotting_2 %>% 
  dabest(neuron, relative_oxidation, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot", 
     effsize.ylim = c(-0.2, .2))

ggsave(here("Plots", "roGFP_5.pdf"))

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c")
```

Comparing axonal vs std (note - very small effect - significant)

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data_for_plotting_2, aes(field, relative_oxidation)) +
  geom_boxplot(width = 0.2, alpha = 0.1, aes(fill = type), position=position_dodge(.3), outlier.alpha = 0)

axonal_only <- data_for_plotting_2 %>% filter(field == "axonal")
std_only <- data_for_plotting_2 %>% filter(field == "axonal")
```

Welch Two Sample t-test - Axonal only

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
t.test(relative_oxidation ~ type, data = axonal_only)
```

Welch Two Sample t-test - STD only

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
t.test(relative_oxidation ~ type, data = std_only)
```

Comparing vulnerable vs non-vulnerable - significant by very small effect size

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data_for_plotting_2, aes(type, relative_oxidation, fill = type)) +
  #geom_point(size = 1, alpha = 0.3, aes(color = field), position = position_jitterdodge(0.05)) +
  geom_boxplot(width = 0.2, alpha = 0.1)

```

Welch Two Sample t-test

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
t.test(relative_oxidation ~ type, data = data_for_plotting_2)
```

\newpage
### Intermitochondrial distance (roGFP)

## Figure 4-3

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment=NA, out.width="50%"}
data <- 
  read_csv(here("derived-data", "derived-mito-distance.csv")) %>% 
  pivot_longer(cols = 1:7, names_to ="neuron", values_to = "distance_um")

data <- 
  data %>%mutate(neuron = fct_relevel(neuron, c("SN", "VTA", "LC", "R", "DMV", "XII", "STR")))

ggplot(data, aes(neuron, distance_um)) +
  theme_classic() +
  geom_boxplot(aes(fill = neuron), width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(color = "black", fill = "black", alpha = 0.1)

ggsave(here("plots", "intermitochondria-distance.pdf"))
```


```{r eval=FALSE, message=FALSE, include=FALSE, paged.print=FALSE}
mean(data$distance_um, na.rm=TRUE)

sd(data$distance_um, na.rm=TRUE)

```

Kurskal-Wallis

```{r 3, echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(distance_um ~ neuron, data = data)

```

Estimation statistics --- not applicable

\newpage
# PD-vulnerable neurons have large axonal domains, that are globally more complex than PD-resilient neurons

## Figure 6-1

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
data <- 
  read_csv(here("derived-data", "trace.csv")) %>% 
  mutate(neuron = fct_relevel(neuron, c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"))) %>% 
  mutate(axon_area_per_neuron = axon_area_per_neuron/1.8274) %>% 
  mutate(type = case_when(
     neuron == "SN" ~ "pd-vulnerable", 
     neuron == "LC" ~ "pd-vulnerable",
     neuron == "R" ~ "pd-vulnerable",
     neuron == "DMV" ~ "pd-vulnerable", 
     neuron == "VTA" ~ "non-vulnerable",
     neuron == "XII" ~ "non-vulnerable", 
     neuron == "STR" ~ "non-vulnerable"))

ggplot(data , aes(neuron, axon_area_per_neuron, fill = neuron)) + 
  geom_jitter(width = 0.005, height = 0.3, size = 2, alpha = 0.1) +
  theme_bw() +
  labs(x = "", y = "mean nueronal surface length (µm)") +
  geom_boxplot(aes(color = neuron), 
               show.legend = FALSE, 
               width = 0.15, 
               alpha = 0.4, 
               position = position_nudge(x = -0.2),
               outlier.alpha = 0) +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4")) +
  scale_y_continuous(limits = c(0, 40000), oob = scales::squish)

ggsave(here("plots", "arbor_per_neuron.pdf"))

```

Kruskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(axon_area_per_neuron ~ neuron, 
             data = data)
```

Dunn

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
dunnTest(axon_area_per_neuron ~ neuron, 
             data = data, method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

Estimation statistics

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data %>%
  dabest(neuron, axon_area_per_neuron, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot")

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c")
```

_PD-vulnerable vs PD-resilient_

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
## for PD vs NON PD

ggplot(data, aes(type, axon_area_per_neuron)) +
  theme_classic() +
  geom_boxplot(width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(data = . %>% filter(type == "non-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  geom_violinhalf(data = . %>% filter(type == "pd-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  labs(x = "", y = "mean nueronal surface length (µm)") +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4")) +
  scale_y_continuous(limits = c(0, 40000), oob = scales::squish)

ggsave(here("plots", "arbor_per_neuron_PD_nonPD.pdf"))

# independent 2-group Mann-Whitney U Test
wilcox.test(data$axon_area_per_neuron~data$type)

two.group.unpaired <- 
  data %>%
  dabest(type, axon_area_per_neuron, 
         # The idx below passes "Control" as the control group, 
         # and "Group1" as the test group. The mean difference
         # will be computed as mean(Group1) - mean(Control1).
         idx = c("non-vulnerable", "pd-vulnerable"), 
         paired = FALSE)

#two.group.unpaired 
two.group.unpaired.meandiff <- mean_diff(two.group.unpaired)
#two.group.unpaired.meandiff

plot(two.group.unpaired.meandiff, color.column = type)
ggsave(here("plots", "arbor_per_neuron_PD_nonPD-Gardner-Altman plot.pdf"))
```

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data, aes(neuron, count_per_neuron, fill = neuron)) + 
  geom_jitter(width = 0.01, height = 0.3, size = 2, alpha = 0.1) +
  theme_bw() +
  labs(x = "", y = "segmentations per neurons") +
  geom_boxplot(aes(color = neuron), 
               show.legend = FALSE, 
               width = 0.15, 
               alpha = 0.4, 
               position = position_nudge(x = -0.2),
               outlier.alpha = 0) +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4"))

ggsave(here("plots", "arbor_segmentation_per_neuron.pdf"))
```

Kruskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(count_per_neuron ~ neuron, 
             data = data)
```

Dunn

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
dunnTest(count_per_neuron ~ neuron, 
             data = data, method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

Estimation statistics

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data %>%
  dabest(neuron, count_per_neuron, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot")

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c")
```

_PD-vulnerable vs PD-resilient_

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data, aes(type, count_per_neuron)) +
  theme_classic() +
  geom_boxplot(width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(data = . %>% filter(type == "non-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  geom_violinhalf(data = . %>% filter(type == "pd-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  labs(x = "", y = "Segmentations per neuron (mean)") +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4")) +
  #scale_y_continuous(limits = c(0, 40000), oob = scales::squish)

ggsave(here("plots", "segmentations_per_neuron_PD_nonPD.pdf"))

# independent 2-group Mann-Whitney U Test
wilcox.test(data$count_per_neuron~data$type)

two.group.unpaired <- 
  data %>%
  dabest(type, count_per_neuron, 
         # The idx below passes "Control" as the control group, 
         # and "Group1" as the test group. The mean difference
         # will be computed as mean(Group1) - mean(Control1).
         idx = c("non-vulnerable", "pd-vulnerable"), 
         paired = FALSE)

#two.group.unpaired 
two.group.unpaired.meandiff <- mean_diff(two.group.unpaired)
#two.group.unpaired.meandiff

plot(two.group.unpaired.meandiff, color.column = type)
ggsave(here("plots", "segmentations_per_neuron_PD_nonPD-Gardner-Altman plot.pdf"))
```

Average segmentation length (µm)

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data, aes(neuron, `Average Size`, fill = neuron)) + 
  geom_jitter(width = 0.01, height = 0.3, size = 2, alpha = 0.1) +
  theme_bw() +
  labs(x = "", y = "Mean segmentation length (µm)") +
  geom_boxplot(aes(color = neuron), 
               show.legend = FALSE, 
               width = 0.15, 
               alpha = 0.4, 
               position = position_nudge(x = -0.2),
               outlier.alpha = 0) +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4"))

ggsave(here("plots", "arbor_average_segmentation_per_neuron.pdf"))
```

Kruskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(`Average Size` ~ neuron, 
             data = data)
```

Dunn

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
dunnTest(`Average Size` ~ neuron, 
             data = data, method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

Estimation statistics

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data %>%
  dabest(neuron, `Average Size`, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot")

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c")
```

_PD-vulnerable vs PD-resilient_

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data, aes(type, `Average Size`)) +
  theme_classic() +
  geom_boxplot(width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(data = . %>% filter(type == "non-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  geom_violinhalf(data = . %>% filter(type == "pd-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  labs(x = "", y = "Segmentations per neuron (mean)") +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4")) 
  #scale_y_continuous(limits = c(0, 40000), oob = scales::squish)

ggsave(here("plots", "arbor_average_segmentation_per_neuron_PD_nonPD.pdf"))
# independent 2-group Mann-Whitney U Test
wilcox.test(data$`Average Size`~data$type)

two.group.unpaired <- 
  data %>%
  dabest(type, `Average Size`, 
         # The idx below passes "Control" as the control group, 
         # and "Group1" as the test group. The mean difference
         # will be computed as mean(Group1) - mean(Control1).
         idx = c("non-vulnerable", "pd-vulnerable"), 
         paired = FALSE)

#two.group.unpaired 
two.group.unpaired.meandiff <- mean_diff(two.group.unpaired)
#two.group.unpaired.meandiff

plot(two.group.unpaired.meandiff, color.column = type)
ggsave(here("plots", "arbor_average_segmentation_per_neuron_PD_nonPD-Gardner-Altman plot.pdf"))

```

\newpage
# PD-vulnerable neurons have a higher proportion of varicosities that are positive for Syt1

## Figure 7-1

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
data_selected <- 
  read_csv(here("derived-data", "derived-varicosity-data.csv")) %>% 
  filter(between(Feret, 2, 25)) %>% 
  filter(between(MinFeret, 2, 25)) %>% 
  filter(AR < 5) %>% 
  mutate(syt_status = case_when(Mean <= 122 ~ "negative", Mean > 122  ~ "positive")) %>% 
  mutate(neuron = fct_relevel(neuron, c("SN", "VTA", "LC", "R", "DMV", "XII", "STR")))

data_percentage_syt1_positive <- 
  data_selected %>% 
  group_by(neuron, image, syt_status) %>% 
  tally() %>% 
  spread(syt_status, n) %>% 
  mutate(percentage_syt1_positive = (positive/(negative+positive) * 100)) %>% 
  ungroup()
```

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
distribution_plot <- 
  ggplot(data_selected, aes(Mean,neuron, fill = neuron)) +
   geom_density_ridges2(jittered_points = TRUE, 
    quantile_lines = TRUE, 
    quantiles = 5, 
    scale = 0.9, 
    alpha = 0.7,
    vline_size = 0.8, 
    vline_color = "red",
    point_size = 0.5, 
    point_alpha = 0.05,
    position = position_raincloud(adjust_vlines = TRUE)) +
  theme_classic() +
  theme(legend.position = "none")

distribution_plot +
  coord_cartesian(xlim = c(0, 500)) +
  xlab("Syt-1 intensity / varicosity") +
  ylab("")

ggsave(here("plots", "varicosites_ridges.pdf"))
```

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
ggplot(data_percentage_syt1_positive, 
       aes(neuron, percentage_syt1_positive)) +
  geom_jitter(width = 0.03, size = 3, alpha = 0.1) +
  theme_bw() +
  labs(x = "", y = "syt-1 positive varicosities (%)") +
  geom_boxplot(aes(color = neuron), width = 0.15, alpha = 0.01, position = position_nudge(x = -0.2)) +
  theme(legend.position = "none") +
  theme_classic()

ggsave(here("plots", "varicosites_percentage_positive.pdf"))
```


Kruskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(percentage_syt1_positive ~ neuron, data = data_percentage_syt1_positive)
```

Dunn

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
dunnTest(percentage_syt1_positive ~ neuron, data = data_percentage_syt1_positive, method="bonferroni")$res %>% 
  knitr::kable(align = "r") 
```

Estimation statistics

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  data_percentage_syt1_positive %>%
  dabest(neuron, percentage_syt1_positive, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot")

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c")
```

_PD-vulnerable vs PD-resilient_

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
data_percentage_syt1_positive <- 
  data_percentage_syt1_positive %>% 
  mutate(type = case_when(
     neuron == "SN" ~ "pd-vulnerable", 
     neuron == "LC" ~ "pd-vulnerable",
     neuron == "R" ~ "pd-vulnerable",
     neuron == "DMV" ~ "pd-vulnerable", 
     neuron == "VTA" ~ "non-vulnerable",
     neuron == "XII" ~ "non-vulnerable", 
     neuron == "STR" ~ "non-vulnerable"))

ggplot(data_percentage_syt1_positive, aes(type, percentage_syt1_positive)) +
  theme_classic() +
  geom_boxplot(width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(data = . %>% filter(type == "non-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  geom_violinhalf(data = . %>% filter(type == "pd-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  labs(x = "", y = "percentage_syt1_positive") +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4")) 
  #scale_y_continuous(limits = c(0, 40000), oob = scales::squish)

ggsave(here("plots", "percentage_syt1_positive_PD_nonPD_PD_nonPD.pdf"))

wilcox.test(data_percentage_syt1_positive$percentage_syt1_positive~data_percentage_syt1_positive$type)

two.group.unpaired <- 
  data_percentage_syt1_positive %>%
  dabest(type, percentage_syt1_positive, 
         # The idx below passes "Control" as the control group, 
         # and "Group1" as the test group. The mean difference
         # will be computed as mean(Group1) - mean(Control1).
         idx = c("non-vulnerable", "pd-vulnerable"), 
         paired = FALSE)

#two.group.unpaired 
two.group.unpaired.meandiff <- mean_diff(two.group.unpaired)
#two.group.unpaired.meandiff

plot(two.group.unpaired.meandiff, color.column = type)
ggsave(here("plots", "percentage_syt1_positive_PD_nonPD-Gardner-Altman plot.pdf"))
```

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
data_selected_nearest_neighbour <- 
  data_selected %>% 
  select(c(neuron, image, X, Y, syt_status, skeleton_axon_pixels, Mean, Area, Feret, MinFeret)) %>% 
  group_by(image) %>% 
  nest() %>% 
  mutate(
    nn = map(data, ~ nndist(.x$X, .x$Y, k = 1)), 
    tidied = map(nn, tidy)) %>%  
  unnest(tidied, data) %>%  
  select(!c(nn, X, Y)) %>% 
  rename(nearest_varicosity = x)

varicosity_nn_length <- 
  left_join(data_percentage_syt1_positive, data_selected_nearest_neighbour) %>% 
  mutate(axon_per_var_px = skeleton_axon_pixels/(positive+negative)) %>% 
  mutate(axon_per_positive_var_px = skeleton_axon_pixels/(positive))

ggplot(varicosity_nn_length, aes((nearest_varicosity/4.83),neuron, fill = neuron)) +
   geom_density_ridges2(jittered_points = TRUE, 
    quantile_lines = TRUE, 
    quantiles = 5, 
    scale = 0.5, 
    alpha = 0.7,
    vline_size = 0.8, 
    vline_color = "red",
    point_size = 0.5, 
    point_alpha = 0.02,
    position = position_raincloud(adjust_vlines = TRUE)) +
  theme_classic() +
  theme(legend.position = "none") +
  xlab(expression(Log[10]~ "nearest varicosity (µm)")) +
  scale_x_log10()
  
ggsave(here("plots", "varicosites_log_nearest_var.pdf"))
```

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
mean_varicosity_nn_length <- 
  varicosity_nn_length %>% 
  select(c(neuron, image, nearest_varicosity)) %>% 
  group_by(neuron, image) %>% 
  summarise(mean_nearest_varicsotiey = mean((nearest_varicosity/4.83)), n = n()) %>% 
  ungroup()

ggplot(mean_varicosity_nn_length, 
       aes(neuron, mean_nearest_varicsotiey)) +
  geom_jitter(width = 0.03, size = 3, alpha = 0.1) +
  theme_bw() +
  labs(x = "", y = "mean nearest varicosity (µm)") +
  geom_boxplot(aes(color = neuron), width = 0.15, alpha = 0.01, position = position_nudge(x = -0.2)) +
  theme(legend.position = "none") +
  theme_classic() +
  coord_cartesian(ylim = c(0, 10))

ggsave(here("plots", "varicosites_nearest_var.pdf"))
```

Kruskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(mean_nearest_varicsotiey ~ neuron, data = mean_varicosity_nn_length)
```

Dunn

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
dunnTest(mean_nearest_varicsotiey ~ neuron, data = mean_varicosity_nn_length, method="bonferroni")$res %>% 
  knitr::kable(align = "r")   
```

Estimation Statistics

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  mean_varicosity_nn_length %>%
  dabest(neuron, mean_nearest_varicsotiey, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot")

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c")
```

_PD-vulnerable vs PD-resilient_

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
mean_varicosity_nn_length <- 
  mean_varicosity_nn_length %>% 
    mutate(type = case_when(
     neuron == "SN" ~ "pd-vulnerable", 
     neuron == "LC" ~ "pd-vulnerable",
     neuron == "R" ~ "pd-vulnerable",
     neuron == "DMV" ~ "pd-vulnerable", 
     neuron == "VTA" ~ "non-vulnerable",
     neuron == "XII" ~ "non-vulnerable", 
     neuron == "STR" ~ "non-vulnerable"))


ggplot(mean_varicosity_nn_length, aes(type, mean_nearest_varicsotiey)) +
  theme_classic() +
  geom_boxplot(width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(data = . %>% filter(type == "non-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  geom_violinhalf(data = . %>% filter(type == "pd-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  labs(x = "", y = "mean_nearest_varicsotiey") +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4")) 
  #scale_y_continuous(limits = c(0, 40000), oob = scales::squish)

ggsave(here("plots", "mean_nearest_varicsotiey_PD_nonPD_PD_nonPD.pdf"))

wilcox.test(mean_varicosity_nn_length$mean_nearest_varicsotiey~mean_varicosity_nn_length$type)

two.group.unpaired <- 
  mean_varicosity_nn_length %>%
  dabest(type, mean_nearest_varicsotiey, 
         # The idx below passes "Control" as the control group, 
         # and "Group1" as the test group. The mean difference
         # will be computed as mean(Group1) - mean(Control1).
         idx = c("non-vulnerable", "pd-vulnerable"), 
         paired = FALSE)

#two.group.unpaired 
two.group.unpaired.meandiff <- mean_diff(two.group.unpaired)
#two.group.unpaired.meandiff

plot(two.group.unpaired.meandiff, color.column = type)
ggsave(here("plots", "varicosites_nearest_var_PD_nonPD-Gardner-Altman plot.pdf"))
```

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
mean_varicosity_lengthpervar_length <- 
  varicosity_nn_length %>% 
  select(c(neuron, image, axon_per_var_px)) %>% 
  group_by(neuron, image) %>% 
  summarise(mean_axon_per_var_px = mean(axon_per_var_px), n = n()) %>% 
  ungroup()
  

ggplot(mean_varicosity_lengthpervar_length, 
       aes(neuron, (mean_axon_per_var_px/4.83))) +
  geom_jitter(width = 0.03, size = 3, alpha = 0.1) +
  theme_bw() +
  labs(x = "", y = "axon length / total varicosities (µm)") +
  geom_boxplot(aes(color = neuron), width = 0.15, alpha = 0.01, position = position_nudge(x = -0.2)) +
  theme(legend.position = "none") +
  theme_classic() + 
 # scale_y_log10()
   coord_cartesian(ylim = c(0, 20))

ggsave(here("plots", "varicosites_var_div_axon.pdf"))
```

Kruskal-Wallis

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
kruskal.test(mean_axon_per_var_px ~ neuron, data = mean_varicosity_lengthpervar_length)
```

Dunn

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
dunnTest(mean_axon_per_var_px ~ neuron, data = mean_varicosity_lengthpervar_length, method="bonferroni")$res %>% 
  knitr::kable(align = "r")  
```

Estimation Statistics

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
shared.control <- 
  mean_varicosity_lengthpervar_length %>%
  dabest(neuron, mean_axon_per_var_px, 
         idx = c("SN", "VTA", "LC", "R", "DMV", "XII", "STR"),
         paired = FALSE
         )

shared.control.mean_diff <- shared.control %>% mean_diff()

plot(shared.control.mean_diff, 
     rawplot.type = "swarmplot")

shared.control.mean_diff$result %>% 
  select(control_group, test_group, difference, bca_ci_low, bca_ci_high) %>% 
  knitr::kable(digits = 3, align = "c")
```

_PD-vulnerable vs PD-resilient_

```{r echo=FALSE, fig.align="center", error=FALSE, message=FALSE, warning=FALSE, comment = NA, out.width="50%"}
mean_varicosity_lengthpervar_length <- 
  mean_varicosity_lengthpervar_length %>% 
    mutate(type = case_when(
     neuron == "SN" ~ "pd-vulnerable", 
     neuron == "LC" ~ "pd-vulnerable",
     neuron == "R" ~ "pd-vulnerable",
     neuron == "DMV" ~ "pd-vulnerable", 
     neuron == "VTA" ~ "non-vulnerable",
     neuron == "XII" ~ "non-vulnerable", 
     neuron == "STR" ~ "non-vulnerable"))

ggplot(mean_varicosity_lengthpervar_length, aes(type, mean_axon_per_var_px)) +
  theme_classic() +
  geom_boxplot(width = 0.2, alpha = 0.1, outlier.alpha = 0, position = position_nudge(x = -0.13)) +
  geom_violinhalf(data = . %>% filter(type == "non-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  geom_violinhalf(data = . %>% filter(type == "pd-vulnerable"), color = NA, alpha = 0.3, aes(fill = type)) +
  labs(x = "", y = "mean_axon_per_var_px") +
  theme_classic() + 
  theme(legend.position = "none", 
        strip.text.x = element_text(size=12, face="bold"),
        strip.background = element_rect(colour="NA", fill="#899DA4")) 
  #scale_y_continuous(limits = c(0, 40000), oob = scales::squish)

ggsave(here("plots", "mean_axon_per_var_px_PD_nonPD_PD_nonPD.pdf"))

wilcox.test(mean_varicosity_lengthpervar_length$mean_axon_per_var_px~mean_varicosity_lengthpervar_length$type)

two.group.unpaired <- 
  mean_varicosity_lengthpervar_length %>%
  dabest(type, mean_axon_per_var_px, 
         # The idx below passes "Control" as the control group, 
         # and "Group1" as the test group. The mean difference
         # will be computed as mean(Group1) - mean(Control1).
         idx = c("non-vulnerable", "pd-vulnerable"), 
         paired = FALSE)

#two.group.unpaired 
two.group.unpaired.meandiff <- mean_diff(two.group.unpaired)
#two.group.unpaired.meandiff

plot(two.group.unpaired.meandiff, color.column = type)
ggsave(here("plots", "varicosites_var_div_axon_PD_nonPD-Gardner-Altman plot.pdf"))
```