---
title: "Statistical analyses"
author: "A.G. West"
date: "25/10/2021"
output: html_document
---

Load packages into session, and print package version.

```{r}
library(knitr)
```


```{r results=F, message=F, warning=F}
library(plyr); packageVersion("plyr")
library(ggplot2); packageVersion("ggplot2")
library(phyloseq); packageVersion("phyloseq")
library(gridExtra); packageVersion("gridExtra")
library(ggsignif); packageVersion("ggsignif")
library(ggsci); packageVersion("ggsci")
library(hrbrthemes); packageVersion("hrbrthemes")
library(tidyverse); packageVersion("tidyverse")
library(viridis); packageVersion("viridis")
library(dplyr); packageVersion("dplyr")
library(extrafont); packageVersion("extrafont")
library(vegan); packageVersion("vegan")
library(ggpubr)

set.seed(100)
```


#Phyloseq

Import files:

```{r}
seqtab.nochim <- readRDS("16S_ASV_table_nochim.rds")
taxa <- readRDS("16S_taxa_table.rds")
tree <- readRDS("phylo_tree.rds")
map <- read.table("mapping_file.txt",header = T, sep = '\t')
row.names(map) <- map$SampleID
  
# Make sure row names are the same between map and ASV table
identical(rownames(seqtab.nochim), rownames(map))
```

```{r}
taxa <- data.frame(taxa)
taxa$Family[is.na(taxa$Family)]="Unclassified"
taxa$Genus[is.na(taxa$Genus)]="Unclassified"
taxa$Species[is.na(taxa$Species)]="unclassified"
taxa$Taxonomy <- paste(taxa$Genus, taxa$Species, sep=" ")

taxmatrix <- as.matrix(taxa) 
```


Create a phyloseq object for downstream analyses:

```{r}
ps <- phyloseq(tax_table(taxmatrix), sample_data(map),otu_table(seqtab.nochim, taxa_are_rows = FALSE), phy_tree(tree$tree))
               
ps
```

#Filtering

##Taxonomic filtering 

Use to filter out non-target taxa e.g. mitochondria, chloroplasts. 

```{r}
dir.create("tables")
# Show available ranks in the dataset
rank_names(ps)

ps_tax_table <- as.data.frame(tax_table(ps))
ps_tax_table$Phylum[is.na(ps_tax_table$Phylum)]="Unclassified"
asv_table_ps<-as.data.frame(t(otu_table(ps)))
asv_table_wTax_ps <- cbind(asv_table_ps, ps_tax_table)

#Find proportion of unassigned reads at phylum level
Phyla.sum.ps<-aggregate(asv_table_wTax_ps[,1:57], list(asv_table_wTax_ps$Phylum),sum)
row.names(Phyla.sum.ps)<- Phyla.sum.ps$Group.1
Phyla.sum.ps$Group.1 <- NULL
P_rowsum.ps <- rowSums(Phyla.sum.ps)
P_rowsum.ps
write.csv(P_rowsum.ps, "tables/P_rowsum_ps.csv")

# Create table, number of features for each phyla
table(tax_table(ps)[, "Phylum"], exclude = NULL)

ps0 <- subset_taxa(ps, !is.na(Phylum) & !Phylum %in% c("", "uncharacterized"))

phyla2Filter = c("<NA>")
ps1 = subset_taxa(ps0, !Phylum %in% phyla2Filter)

phyla2Filter = c("Chloroplast","Rickettsiales")
ps1 = subset_taxa(ps1, !Order %in% phyla2Filter)

ps1
rank_names(ps1)
table(tax_table(ps1)[,"Phylum"], exclude = NULL)
```


##Filter low-abundance ASVs

Filter out ASVs that have a total relative sequence abundance of < 0.001%

```{r}
minTotRelAbun = 1e-5 #(0.00001)
x = taxa_sums(ps1)
keepTaxa = (x / sum(x)) > minTotRelAbun  
ps2 = prune_taxa(keepTaxa, ps1)
ps2

# Save the new taxa table for rarefaction
Ntaxa <- tax_table(ps2)
```

#Rarefaction


```{r}
set.seed(100)
rASV <- rarefy_even_depth(otu_table(ps2, taxa_are_rows = FALSE), sample.size = 8000,replace = FALSE) 

#Save the rarefied ASV table
write.csv(rASV, "./tables/rASV_table.csv")
```

Relabel ASV sequence rownames with 'ASV1_Taxonomy'

```{r}
rASV.df = as.data.frame(rASV)
rASV.df = rASV.df[,order(colSums(rASV.df),decreasing = T)]
Ntaxa.df = as.data.frame(Ntaxa)

to.remove = setdiff(rownames(Ntaxa.df), colnames(rASV.df))
Ntaxa.df = Ntaxa.df[!row.names(Ntaxa.df) %in% to.remove,]

identical(rownames(Ntaxa.df), colnames(rASV.df))
Ntaxa.df = Ntaxa.df[colnames(rASV.df),]
identical(rownames(Ntaxa.df), colnames(rASV.df))

Ntaxa.df$ASV_ID <- paste("ASV_", 1:nrow(Ntaxa.df), sep="")
Ntaxa.df$concat = paste(Ntaxa.df$ASV_ID, Ntaxa.df$Taxonomy, sep = "_")
rownames(Ntaxa.df) = Ntaxa.df$concat
names(rASV.df) = rownames(Ntaxa.df)

Ntaxa.df$ASV_ID = NULL
Ntaxa.df$concat = NULL
Ntaxa.df = as.matrix(Ntaxa.df)
```

Create rarefied phyloseq objects - cannot create phyloseq object with the tree file after renaming ASVs so here we will create two seperate objects.

```{r}
psR <- phyloseq(otu_table(rASV.df, taxa_are_rows = F),
                sample_data(map),
                tax_table(Ntaxa.df))

psR

write.csv(cbind(t(rASV),as.data.frame(tax_table(psR))), "tables/rASV_wTax.csv")
```

```{r}
psRT <- phyloseq(otu_table(rASV, taxa_are_rows = F),
                 sample_data(map),
                 tax_table(Ntaxa),
                 phy_tree(tree$tree))

psRT
```

#Taxonomy sums

Calculate the number of ASVs per sample and the prevalence of each ASV

```{r}
asv_df <- t(otu_table(psR))
avg_ASV <- colSums(asv_df != 0)
write.csv(avg_ASV, "./tables/avg_ASV.csv")

prev_asv_df <- otu_table(psR)
prev_ASV <- colSums(prev_asv_df != 0)
write.csv(prev_ASV, "./tables/prev_ASV.csv") 
```

Find the total number of reads per sample and totals for phylum and species level classifications

```{r}
asv_table<-as.data.frame(t(otu_table(psR)))
taxo_table<-as.data.frame(tax_table(psR))
asv_table_wTax <- cbind(asv_table, taxo_table)

Phyla.sum<-aggregate(asv_table_wTax[,1:57], list(asv_table_wTax$Phylum),sum)
row.names(Phyla.sum)<- Phyla.sum$Group.1
Phyla.sum$Group.1 <- NULL
P_rowsum <- rowSums(Phyla.sum)

write.csv(Phyla.sum, "./tables/Phyla_sum_per_sample.csv")
write.csv(P_rowsum, "./tables/P_rowsum.csv")
```

```{r}
taxa.sum <- aggregate(asv_table_wTax[,1:57], list(asv_table_wTax$Taxonomy),sum)
row.names(taxa.sum)<- taxa.sum$Group.1
taxa.sum$Group.1 <- NULL
t_rowsum <- rowSums(taxa.sum) 
write.csv(t_rowsum, "./tables/taxa_sum_table.csv")
write.csv(taxa.sum, "./tables/taxa_sum_per_sample.csv")
```

#Core Microbiome

```{r  results=F, message=F, warning=F}
library(microbiome); packageVersion("microbiome")
```


```{r fig.cap="Core taxa of the takah\u113 gut microbiota", fig.height = 15, fig.width = 22, dpi=300}
pseq.rel <- microbiome::transform(psR, "compositional")

#Core at >0.1% abundance and 90% prevalence
core.taxa.standard <- core_members(pseq.rel, detection = 0.1/100, prevalence = 90/100)
core.taxa.standard
```

```{r}
#Core at >0.01% abundance and 70% prevalence
core.taxa.standard <- core_members(pseq.rel, detection = 0.01/100, prevalence = 70/100) #pick the core
core.taxa.standard

#Relative population frequencies
head(prevalence(psR, detection = 1/100, sort = TRUE, count = T, include.lowest = T), n=20) #absolute counts
head(prevalence(pseq.rel, detection = 0.01/100, sort = TRUE, include.lowest = T), n=20) #relative prevalence
```

```{r}
pseq.core <- core(pseq.rel, detection = 0.01/100, prevalence = 70/100)

core_RA <- transform_sample_counts(pseq.core, function(x) {x/sum(x)})
core.glom <- tax_glom(core_RA, taxrank = 'Taxonomy')
core.df <- psmelt(core.glom)

core.glom.absolute <- tax_glom(pseq.core, taxrank = "Taxonomy")
core.ab.df <- psmelt(core.glom.absolute)

dir.create("core")
```

```{r fig.cap = "Core microbiota of the takah\u113 gut grouped by location", fig.height = 15, fig.width = 22, dpi=300}
core_plot <- ggplot(data=core.df, aes(x=Actual_Name, y=Abundance, fill = Taxonomy)) + 

  geom_bar(aes(), stat="identity", position="stack")  +
  
  theme_ipsum() + theme(axis.text.x = element_blank(),

          axis.ticks.x = element_blank(),

          axis.text.y=element_text(size=15),

          axis.title.y=element_text(size=26),
          
          axis.title.x = element_text(size=26),
          
          strip.text.x = element_text(size = 18),

          panel.grid.major = element_blank(),

          panel.grid.minor = element_blank(),

          panel.background = element_blank(),

          legend.box.background = element_rect(),

          legend.box.margin = margin(5, 5, 5, 5),

          legend.position="bottom",

          legend.title = element_text(face = "bold", size = 24),
          
          legend.text = element_text(size = 20)) +
  
  labs(x="Individual faecal samples", y="Relative sequence abundance") + 
  
  facet_grid(~Location, scales = "free_x", space = "free",labeller = labeller(Location = label_wrap_gen(10))) +
  
  scale_fill_futurama(alpha=0.85) 

core_plot

ggsave("./core/core_microbiome_taxaplot.png",height = 15, width = 22, dpi=400, bg = "white") 
```
```{r fig.cap = "Proportion of core taxa in the takah\u113 gut grouped by location", fig.height = 15, fig.width = 22, dpi=300}
core_absolute_plot <- ggplot(data=core.ab.df, aes(x=Actual_Name, y=Abundance, fill = Taxonomy)) + 

  geom_bar(aes(), stat="identity", position="stack")  +
  
  theme_ipsum() + theme(axis.text.x = element_blank(),

          axis.ticks.x = element_blank(),

          axis.text.y=element_text(size=15),

          axis.title.y=element_text(size=26),
          
          axis.title.x = element_text(size=26),
          
          strip.text.x = element_text(size = 18),

          panel.grid.major = element_blank(),

          panel.grid.minor = element_blank(),

          panel.background = element_blank(),

          legend.box.background = element_rect(),

          legend.box.margin = margin(5, 5, 5, 5),

          legend.position="bottom",

          legend.title = element_text(face = "bold", size = 24),
          
          legend.text = element_text(size = 20)) +
  
 labs(x="Individual faecal samples", y="Relative sequence abundance") + 
  
  facet_grid(~Location, scales = "free_x", space = "free",labeller = labeller(Location = label_wrap_gen(10))) +
  
  scale_fill_futurama(alpha=0.85) 

core_absolute_plot

ggsave("./core/core_abosulte_microbiome_taxaplot.png", height = 15, width = 22,dpi=400, bg = "white") 
```

#Alpha Diversity

Use phyloseq's estimate_richness function to get desired diversity estimates. Test for normality and then employ ANOVA or Kruskal-Wallis tests for each target covariate.


```{r}
rich_psR <- estimate_richness(psR, measures = c("Observed","Shannon", "InvSimpson"))
rich <- list(rich_psR, sample_data(psR)$SampleID, sample_data(psR)$Location, sample_data(psR)$Supplemental_feeding, sample_data(psR)$Hatch_Site, sample_data(psR)$Nest_Site, sample_data(psR)$Sex, sample_data(psR)$Age, sample_data(psR)$Natal.Year, sample_data(psR)$Habitat, sample_data(psR)$Origin) 
names(rich) <- c("alpha_diversity", "SampleID", "Location","Supplemental feeding", "Hatch_Site", "Nest_Site", "Sex", "Age", "Natal_Year", "Habitat", "Origin")
```

```{r, results='hide'}
# Location
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$Location, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Observed, rich$Location))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$Location, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Shannon, rich$Location))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$Location, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$InvSimpson, rich$Location))}

# Supplemental Feeding
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$`Supplemental feeding`, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Observed, rich$`Supplemental feeding`))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$`Supplemental feeding`, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Shannon, rich$`Supplemental feeding`))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$`Supplemental feeding`, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$InvSimpson, rich$`Supplemental feeding`))}

# Hatch Site
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$Hatch_Site, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Observed, rich$Hatch_Site))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$Hatch_Site, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Shannon, rich$Hatch_Site))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$Hatch_Site, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$InvSimpson, rich$Hatch_Site))}

# Nest Site
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$Nest_Site, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Observed, rich$Nest_Site))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$Nest_Site, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Shannon, rich$Nest_Site))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$Nest_Site, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$InvSimpson, rich$Nest_Site))}

# Sex
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$Sex, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Observed, rich$Sex))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$Sex, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Shannon, rich$Sex))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$Sex, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$InvSimpson, rich$Sex))}

# Age
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$Age, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Observed, rich$Age))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$Age, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Shannon, rich$Age))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$Age, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$InvSimpson, rich$Age))}

# Natal year
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$Natal_Year, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Observed, rich$Natal_Year))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$Natal_Year, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$Shannon, rich$Natal_Year))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$Natal_Year, data = rich)))} else {print(kruskal.test(rich$alpha_diversity$InvSimpson, rich$Natal_Year))}

# Habitat type
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(t.test(rich$alpha_diversity[["Observed"]]~rich$Habitat, data = rich))} else {print(wilcox.test(alpha_diversity$Observed ~ Habitat, data=rich))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(t.test(rich$alpha_diversity[["Shannon"]]~rich$Habitat, data = rich))} else {print(wilcox.test(alpha_diversity$Shannon ~ Habitat, data=rich))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(t.test(rich$alpha_diversity[["InvSimpson"]]~rich$Habitat, data = rich))} else {print(wilcox.test(alpha_diversity$InvSimpson ~ Habitat, data=rich))}

# Origin
if (shapiro.test(rich$alpha_diversity[["Observed"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Observed"]]~rich$Origin, data = rich)))} else {print(wilcox.test(alpha_diversity$Observed ~ Origin, data=rich))}

if (shapiro.test(rich$alpha_diversity[["Shannon"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["Shannon"]]~rich$Origin, data = rich)))} else {print(kruskal.test(alpha_diversity$Shannon ~ Origin, data=rich))}

if (shapiro.test(rich$alpha_diversity[["InvSimpson"]])$p.value > 0.05)
{print(summary(aov(rich$alpha_diversity[["InvSimpson"]]~rich$Origin, data = rich)))} else {print(kruskal.test(alpha_diversity$InvSimpson ~ Origin, data=rich))}
```

Pairwise testing for significant results using Dunn's test with Benjamini-Hochberg adjustment

```{r  results=F, message=F, warning=F}
library(dunn.test); packageVersion("dunn.test")
```
```{r}
dunn.test(rich$alpha_diversity$Shannon, rich$Location, method = "bh")

dunn.test(rich$alpha_diversity$InvSimpson, rich$Location, method = "bh")

dunn.test(rich$alpha_diversity$InvSimpson, rich$`Supplemental feeding`, method = "bh")
```

Plot alpha-diversity

```{r}
library(Manu); packageVersion("Manu")
takahe_pal <- c("#DD3C51", "#313657", "#51806a","#1F6683", "#83A552","#6C90B9","#E2939E", "#D1C7B5")

#p-values from Dunn's tests
annotation_df_shannon <- data.frame(start=c("Burwood Centre", "Burwood Centre","Burwood Centre"), 
                            end=c("Cape Sanctuary", "Foveaux Strait Is", "Murchison Mountain"),
                            y=c(4.6, 5.1,5.6),
                            label=c("0.03*", "0.02*","0.04*"))

annotation_df_simpson <- data.frame(start=c("Burwood Centre", "Burwood Centre","Burwood Centre", "Burwood Centre"), 
                            end=c("Cape Sanctuary", "Foveaux Strait Is", "Murchison Mountain", "Tiritiri Matangi Is"),
                            y=c(34,40,46,52),
                            label=c("0.04*", "0.02*","0.008**","0.03*"))


myadptheme <- theme_ipsum() + theme(plot.title = element_text(size = 26),
  
          plot.subtitle = element_text(size = 24),
          
          axis.text.x = element_blank(),

          axis.title.x = element_text(size=35),

          axis.ticks.x = element_blank(),

          axis.text.y=element_text(size=25),

          axis.title.y=element_text(size=35),

          axis.line.x=element_line(color="black",size=1.0,linetype=1),

          axis.line.y=element_line(color="black",size=1.0,linetype=1),

          panel.grid.major = element_blank(),

          panel.grid.minor = element_blank(),

          panel.background = element_blank(),

          legend.box.background = element_rect(),

          legend.box.margin = margin(5, 5, 5, 5),

          legend.position="bottom",

          legend.title = element_text(face = "bold", size = 35),
          
          legend.text = element_text(size = 30))

rich_df <- as.data.frame(rich)
names(rich_df)[names(rich_df) == "alpha_diversity.Observed"] <- "Observed"
names(rich_df)[names(rich_df) == "alpha_diversity.Shannon"] <- "Shannon"
names(rich_df)[names(rich_df) == "alpha_diversity.InvSimpson"] <- "InvSimpson"
```

###Location
```{r fig.cap="Alpha diversity of the takah\u113 gut microbiota",  fig.height = 10, fig.width = 15, dpi=300}
Observed <- ggplot(rich_df, aes(x=Location, y=Observed))  
Observed.plot <- Observed + geom_boxplot(size=1.0, aes(fill=Location)) +   
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_fill_manual(values = takahe_pal) +
            labs(y = "Observed richness", x = "Location") +
            ggtitle("ANOVA p = 0.14")
Observed.plot 


shannon <- ggplot(rich_df, aes(x=Location, y=Shannon))  
shannon.plot <- shannon + geom_boxplot(size=1.0, aes(fill=Location)) +   
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_y_continuous(limits = c(0,6)) +
            scale_fill_manual(values = takahe_pal) +
            labs(y = "Shannon diversity", x = "Location") +
            ggtitle("Kruskal-Wallis p = 0.01*") + 
            geom_signif(data=annotation_df_shannon,aes(xmin=start, xmax=end, annotations=label, y_position=y),
              manual=TRUE, tip_length = 0, textsize = 8, colour = "blue") 
shannon.plot


simpson <- ggplot(rich_df, aes(x=Location, y=InvSimpson))
simpson.plot <- simpson + geom_boxplot(size=1.0, aes(fill=Location)) +   
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_fill_manual(values = takahe_pal) +
            labs(y = "Inverse Simpson diversity", x = "Location") +
            ggtitle("Kruskal-Wallis p = 0.007**") + 
            geom_signif(data=annotation_df_simpson,aes(xmin=start, xmax=end, annotations=label, y_position=y),
              manual=TRUE, tip_length = 0, textsize = 8, colour = "blue") 
simpson.plot
```

```{r fig.height=10, fig.width=25}
library(ggpubr)
location.adp <- ggarrange(Observed.plot, shannon.plot, simpson.plot, labels = c("A"), font.label = list(size = 30),
          ncol=3, nrow=1, common.legend = T , legend = "bottom")

location.adp
```

###Supplemental Feeding

```{r fig.cap = "Alpha Diversity measures in the takah\u113 gut microbiota by supplemental feeding", fig.height = 10, fig.width = 15, dpi=300}
annoSF_simpson <- data.frame(start=c("None"), 
                            end=c("Regular"),
                            y=c(50),
                            label=c("0.003**"))

f_eight <- c("#FF6F00BF", "#C71000BF", "#008EA0BF", "#8A4198BF", "#FF6348BF", "#5A9599BF","#84D7E1BF", "#FF95A8BF")


ObservedSF <- ggplot(rich_df, aes(x=Supplemental.feeding, y=Observed))  
ObservedSF.plot <- ObservedSF + geom_boxplot(size=1.0, aes(fill=Supplemental.feeding)) +   
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_fill_manual(values = f_eight, name = "Supplemental feeding") +
            labs(y = "Observed richness", x = "Supplemental feeding") +
            ggtitle("ANOVA p = 0.75")
ObservedSF.plot


shannonSF <- ggplot(rich_df, aes(x=Supplemental.feeding, y=Shannon))  
shannonSF.plot <- shannonSF + geom_boxplot(size=1.0, aes(fill=Supplemental.feeding)) + 
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_fill_manual(values = f_eight, name = "Supplemental feeding") +
            labs(y = "Shannon diversity", x = "Supplemental feeding") +
            ggtitle("Kruskal-Wallis p = 0.06")
shannonSF.plot

simpsonSF <- ggplot(rich_df, aes(x=Supplemental.feeding, y=InvSimpson))  
simpsonSF.plot <- simpsonSF + geom_boxplot(size=1.0, aes(fill=Supplemental.feeding)) +  
              myadptheme + theme(plot.title = element_text(size = 35)) +
              scale_fill_manual(values = f_eight, name = "Supplemental feeding") +
              labs(y = "Inverse Simpson diversity", x = "Supplemental feeding") +
              ggtitle("Kruskal-Wallis p = 0.009**") +
              geom_signif(data=annoSF_simpson,aes(xmin=start, xmax=end, annotations=label, y_position=y),
              manual=TRUE, tip_length = 0, textsize = 8, colour = "blue") 
simpsonSF.plot

```

```{r fig.height=10, fig.width=25}
SF.adp <- ggarrange(ObservedSF.plot, shannonSF.plot, simpsonSF.plot, labels = c("B"), font.label = list(size = 30),
          ncol=3, nrow=1, common.legend = T , legend = "bottom")

SF.adp
```

###Habitat
```{r fig.cap = "Alpha Diversity measures in the takah\u113 gut microbiota by habitat type", fig.height = 10, fig.width = 15, dpi=300}
Observed.H <- ggplot(rich_df, aes(x=Habitat, y=Observed))  
Observed.H.plot <- Observed.H + geom_boxplot(size=1.0, aes(fill=Habitat)) +   
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_fill_manual(values = c("#5A9599BF","#84D7E1BF")) +
            labs(y = "Observed richness", x = "Habitat") +
            ggtitle("ANOVA p = 0.005**")

shannon.H <- ggplot(rich_df, aes(x=Habitat, y=Shannon))  
shannon.H.plot <- shannon.H + geom_boxplot(size=1.0, aes(fill=Habitat)) +   
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_fill_manual(values = c("#5A9599BF","#84D7E1BF")) +
            labs(y = "Shannon diversity", x = "Habitat") +
            ggtitle("Wilcoxon p = 0.05")


simpson.H <- ggplot(rich_df, aes(x=Habitat, y=InvSimpson))  
simpson.H.plot <- simpson.H + geom_boxplot(size=1.0, aes(fill=Habitat)) +   
            myadptheme + theme(plot.title = element_text(size = 35)) +
            scale_fill_manual(values = c("#5A9599BF","#84D7E1BF")) +
            labs(y = "Inverse Simpson diversity", x = "Habitat") +
            ggtitle("Wilcoxon p = 0.1")
```

```{r fig.height=10, fig.width=25}
Habitat.adp <- ggarrange(Observed.H.plot, shannon.H.plot, simpson.H.plot, labels = c("C"), font.label = list(size = 30),
          ncol=3, nrow=1, common.legend = T , legend = "bottom")

Habitat.adp
```

```{r results=F, message=F, warning=F}
library(cowplot)
```

```{r fig.height = 30, fig.width = 25}
adp_all_together = ggarrange(location.adp, NULL, SF.adp, NULL, Habitat.adp, ncol = 1, nrow = 5, heights = c(1,0.2,1,0.2,1))
adp_all_together

ggsave("alphadiv_together.png", adp_all_together, height = 30, width = 25, dpi = 400, bg= "white")
```


#Taxa plots

Use plyr to create an 'Others' group based on mean abundance.

```{r results=F, message=F, warning=F}
library(plyr); packageVersion("plyr")
```


```{r}
others <- transform_sample_counts(psR, function(x) x/sum(x))

# agglomerate taxa
glom <- tax_glom(others, taxrank = 'Phylum')

# create dataframe from phyloseq object
df <- psmelt(glom)

# convert Phylum to a character vector
df$Phylum <- as.character(df$Phylum)

# group dataframe by Phylum, calculate mean rel. abundance
means <- ddply(df, ~Phylum, function(x) c(mean=mean(x$Abundance)))

st.dev <- ddply(df, ~Phylum, function(x) c(std=sd(x$Abundance)))

# The code is creating a list of mean values for the condensed taxonomies across the data set.
# It does not effect the actual abundances of each ASV within the taxa group (which is what you plot)
# The list of mean values is then used to rename taxa whose mean is less than the value specified e.g. 0.001

# find Phyla whose rel. abund. is less than 1%
remainder <- means[means$mean <= 0.001,]$Phylum 

# change name to "Others"
df[df$Phylum %in% remainder,]$Phylum <- 'Other phyla <0.1%' 

#Find how many colours are required
print(levels(factor(df$Phylum)))
```

```{r}
dir.create("taxaplots")

taxa_theme <- theme_ipsum() + theme(axis.text.x = element_blank(),

          axis.ticks.x = element_blank(),

          axis.text.y=element_text(size=15),

          axis.title.y=element_text(size=26),
          
          axis.title.x = element_text(size=26),
          
          strip.text.x = element_text(size = 22),

          panel.grid.major = element_blank(),

          panel.grid.minor = element_blank(),

          panel.background = element_blank(),

          legend.box.background = element_rect(),

          legend.box.margin = margin(5, 5, 5, 5),

          legend.position="bottom",

          legend.title = element_text(face = "bold", size = 30),
          
          legend.text = element_text(size = 25))
```


```{r fig.cap = "Phylum-level taxonomy plot of the takah\u113 gut microbiota grouped by location", fig.height = 15, fig.width = 22, dpi=300}
POplot <- ggplot(data=df, aes(x=Actual_Name, y=Abundance, fill = factor(Phylum, levels =  c("Acidobacteriota","Actinobacteriota","Bacteroidota","Campilobacterota","Firmicutes","Fusobacteriota","Planctomycetota","Proteobacteria","Verrucomicrobiota","Other phyla <0.1%"))))

phylum_plot <- POplot + geom_bar(aes(), stat="identity", position="stack") + 
  
  guides(fill=guide_legend(title="Phylum")) +
  
  scale_fill_futurama(alpha=0.75) +
  
  taxa_theme +
  
  labs(x="Individual faecal samples", y="Relative sequence abundance") + 
  
  facet_grid(~Location, scales = "free_x", space = "free" ,labeller = labeller(Location = label_wrap_gen(11)))
   
phylum_plot
ggsave("./taxaplots/phylum_plot.png", plot = phylum_plot, width = 22, height = 15, dpi = 400, bg = "white")
```

By Supplemental feeding
```{r fig.cap = "Phylum-level taxonomy plot of the takah\u113 gut microbiota grouped by supplemental feeding", fig.height = 15, fig.width = 22, dpi=300}
phylum_plotSF <- POplot + geom_bar(aes(), stat="identity", position="stack")  + 
  
  guides(fill=guide_legend(title="Phylum")) +

  taxa_theme +  labs(x="Inidividual faecal sample", y="Relative sequence abundance") + 
  
  facet_grid(~Supplemental_feeding, scales = "free_x", space = "free") +
  
  scale_fill_futurama(alpha=0.75) 

phylum_plotSF
ggsave("./taxaplots/phylum_plotSF.png", plot = phylum_plotSF, width = 22, height = 15, dpi = 400, bg = "white")
```

Genus level taxonomic plots.

```{r}
glomG <- tax_glom(others, taxrank = 'Genus')

dfG <- psmelt(glomG)

dfG$Genus <- as.character(dfG$Genus)

meansG <- ddply(dfG, ~Genus, function(x) c(mean=mean(x$Abundance)))

remainderG <- meansG[meansG$mean <= 0.01,]$Genus

dfG[dfG$Genus %in% remainderG,]$Genus <- 'Other genera <1%'

print(levels(factor(dfG$Genus)))
```

```{r fig.cap = "Genus-level taxonomic plot of the takah\u113 gut microbiota", fig.height = 15, fig.width = 22, dpi=300}
taxa.colours.list.2021 <-  c("#882E72","#B178A6","#D6C1DE","#1965B0","#5289C7","#7BAFDE","#4EB265","#CAE0AB","#F7EE55","#F6C141","#E8601C","#DC050C","#FF0F39","#FF99AB","#ECD3D8","#B1B1B1","#767676","#54809D","#77b6b1","#33605d")

GOplot <- ggplot(data=dfG, aes(x=Actual_Name, y=Abundance,  fill = factor(Genus, levels =  c("Actinobacillus","Anaerobiospirillum","Bacteroides","Bifidobacterium","Campylobacter","Catellicoccus","Clostridium sensu stricto 1","Dialister","Escherichia-Shigella","Fusobacterium","Lachnospiraceae NK3A20 group","Lactobacillus","Megamonas","Prevotella","Pseudomonas","Solobacterium","Sutterella","Turicibacter","Ureaplasma","Other genera <1%"))))


Genus_plot <- GOplot + geom_bar(aes(), stat="identity", position="fill")  + 
  
  guides(fill=guide_legend(title="Genus")) + 
  
  taxa_theme +
  
  labs(x="Individual faecal samples", y="Relative sequence abundance") +  
  
  facet_grid(~Location, scales = "free_x", space = "free" ,labeller = labeller(Location = label_wrap_gen(11))) +
  
  scale_fill_manual(values = taxa.colours.list.2021)

Genus_plot
ggsave("./taxaplots/Genus_taxa_plot_nonsorted.png", height = 15, width = 22, dpi=400, bg = "white")
```

Reorder samples by Lactobacillus abundance


```{r results=F, message=F, warning=F}
library(dplyr)
library(tidyverse); packageVersion("tidyverse")
```


```{r, results='hide'}
dfG %>% 
  group_by(SampleID) %>% 
  summarise(Sum = sum(Abundance)) %>% 
  arrange(Sum) %>% 
  print(n = Inf)

library(forcats); packageVersion("forcats")

dfG_reorder <- dfG 

dfG_reorder$SampleID_new <- 
  as.character(dfG_reorder$SampleID)

SampleID_new_levels <-
  dfG_reorder %>%
  filter(Genus == "Lactobacillus")  %>%
  group_by(SampleID_new)%>% 
  summarise(Sum = sum(Abundance)) %>%
  arrange(Sum) %>%
  pull(SampleID_new) %>%
  unique

dfG_reorder$SampleID_new <-
  factor(dfG_reorder$SampleID_new,
         levels = SampleID_new_levels)
```

```{r results=F, message=F, warning=F}
library(phyloseq)
library(ggplot2)
library(hrbrthemes)
```


```{r fig.cap = "Genus-level taxonomic plot ordered by relative abundance of Lactobacillus", fig.height = 15, fig.width = 22, dpi=300}
LactoGOplot <- ggplot(data=dfG_reorder, aes(x=SampleID_new, y=Abundance, fill = factor(Genus, levels =  c("Actinobacillus","Anaerobiospirillum","Bacteroides","Bifidobacterium","Campylobacter","Catellicoccus","Clostridium sensu stricto 1","Dialister","Escherichia-Shigella","Fusobacterium","Lachnospiraceae NK3A20 group","Lactobacillus","Megamonas","Prevotella","Pseudomonas","Solobacterium","Sutterella","Turicibacter","Ureaplasma","Other genera <1%"))))

LactoGO_plot <- LactoGOplot + geom_bar(aes(), stat="identity", position="fill") +
  
  guides(fill=guide_legend(title="Genus")) + 
  
  taxa_theme +
  
  labs(x="Individual faecal samples", y="Relative sequence abundance") +  
  
  facet_grid(~Location, scales = "free_x", space = "free",labeller = labeller(Location = label_wrap_gen(11))) +
  
  scale_fill_manual(values = taxa.colours.list.2021)

LactoGO_plot
ggsave("./taxaplots/LactoGO_plot.png",width = 22,height = 15, dpi = 400, bg = "white")
```

By Supplemental feeding

```{r fig.cap = "Genus-level taxonomic plot ordered by relative abundance of Lactobacillus grouped by supplemental feeding", fig.height = 12, fig.width = 22, dpi=300}
LactoGO_SF_plot <- LactoGOplot + geom_bar(aes(), stat="identity", position="fill")  + 
  
  guides(fill=guide_legend(title="Genus")) + 
  
  taxa_theme +
  
  labs(x="Individual faecal samples", y="Relative sequence abundance") +  
  
  facet_grid(~Supplemental_feeding, scales = "free_x", space = "free") +
  
  scale_fill_manual(values = taxa.colours.list.2021)


LactoGO_SF_plot
ggsave("./taxaplots/LactoGO_SF_plot.png",width = 22,height = 15, dpi=400, bg = "white")
```

```{r fig.height=25, fig.width=18}
library(cowplot)
taxa_together <- plot_grid(phylum_plot, Genus_plot, labels = c("A","B"), label_size = 30, ncol=1)
taxa_together
ggsave("taxa_together.png", taxa_together, height = 25, width = 24, dpi=400)
```



#Beta Diversity

##BrayCurtis nMDS
```{r}
library(ggtext)

ord_theme = theme_ipsum() + 
  theme(plot.title = element_markdown(size = 35, margin=margin(0,0,30,0)),
    
    axis.title.x = element_text(size=35),
    
    axis.title.y =element_text(size=35),

    axis.line.x=element_line(color="black",size=0.5,linetype=1),

    axis.line.y=element_line(color="black",size=0.5,linetype=1),

    panel.grid.major = element_blank(),

    panel.grid.minor = element_blank(),

    panel.background = element_blank(),

    legend.text = element_markdown(size=25),
    
    legend.title = element_markdown(face="bold", size = 30),
    
    legend.position = "top") 
```


```{r}
library(vegan)
psR_RA <- transform_sample_counts(psRT, function(x) {x/sum(x)})
otu_RA <- data.frame(otu_table(psR_RA))
psRA.dist <- vegdist(otu_RA, method="bray",
                    binary=FALSE, diag=FALSE, upper=FALSE,
                    na.rm=FALSE)

bray.nmds.ord <- metaMDS(psRA.dist)

bray.nmds.ord #0.17 stress

nmds.data <- as.data.frame(scores(bray.nmds.ord))

nmds.data$Location <- map$Location
nmds.data$Supplemental_feeding <- map$Supplemental_feeding
nmds.data$Habitat <- map$Habitat
nmds.data$Lactobacillus <- map$Lactobacillus
```

```{r fig.cap = "Bray Curtis PCoA plot on takah\u113 ASV sequence data by location and Lactobacillus abundance", fig.height = 12, fig.width = 18, dpi=300}
dir.create("ordinations")

nmds.lactob <- ggplot() +
  geom_point(data=nmds.data, aes(x = NMDS1, y = NMDS2, color = Lactobacillus, shape=Location), size=7, stroke = 2) +
  geom_vline(xintercept = 0, linetype = "dashed") + 
  geom_hline(yintercept = 0, linetype = "dashed") +
  scale_shape_manual(values=c(15:17,21:25)) +
  ord_theme + theme(legend.box = "vertical") + 
  guides(shape = guide_legend(order = 1)) +
  scale_color_gradient(low = "blue", high = "red", name="*Lactobacillus* abundance",
                       labels=c("Low","High"),breaks=c(1000,5300), guide = "colourbar") +
 
  ggtitle("Bacterial communities by *Lactobacillus* abundance")

nmds.lactob
ggsave("./ordinations/bray_pcoa_plot.png", width = 18, height = 12, dpi = 400, bg = "white")
```


```{r fig.cap = "Bray Curtis PCoA plot on takah\u113 ASV sequence data by supplemental feeding", fig.height = 12, fig.width = 18, dpi=300}
f_eight <- c("#FF6F00BF", "#C71000BF", "#008EA0BF", "#8A4198BF", "#FF6348BF", "#5A9599BF","#84D7E1BF", "#FF95A8BF")

nmds.SF <- ggplot() +
  geom_point(data=nmds.data, aes(x = NMDS1, y = NMDS2, color = Supplemental_feeding, shape=Location), size=7, stroke = 2) +
  geom_vline(xintercept = 0, linetype = "dashed") + 
  geom_hline(yintercept = 0, linetype = "dashed") +
  scale_shape_manual(values=c(15:17,21:25)) +
  ord_theme + theme(legend.box = "vertical") + 
  guides(shape = guide_legend(order = 1)) +
  scale_color_manual(values = f_eight, name="Supplemental feeding") +
  ggtitle("Bacterial communities by supplemental feeding")

nmds.SF
```


```{r fig.cap = "Bray Curtis PCoA plot on takah\u113 ASV sequence data by habitat type", fig.height = 12, fig.width = 18, dpi=300}
nmds.Habitat <- ggplot() +
  geom_point(data=nmds.data, aes(x = NMDS1, y = NMDS2, color = Habitat, shape=Location), size=7, stroke = 2) +
  geom_vline(xintercept = 0, linetype = "dashed") + 
  geom_hline(yintercept = 0, linetype = "dashed") +
  scale_shape_manual(values=c(15:17,21:25)) +
  ord_theme + theme(legend.box = "vertical") + 
  guides(shape = guide_legend(order = 1)) +
  scale_color_manual(values = c("#5A9599BF","#84D7E1BF"), name="Habitat") +
  ggtitle("Bacterial communities by habitat")

nmds.Habitat
```

```{r fig.height=25, fig.width=18}
library(cowplot)
nmds.SF.HB <- plot_grid(nmds.SF, nmds.Habitat, ncol = 1, nrow = 2, labels = c("A","B"), label_size = 30)
nmds.SF.HB

ggsave("./ordinations/nmds_sf_hb_together.jpg", nmds.SF.HB, height = 25, width = 18, dpi = 300, bg = "white")
```


##gUniFrac PCoA

```{r results=F, message=F, warning=F}
library(GUniFrac); packageVersion("GUniFrac")
library(phangorn); packageVersion("phangorn")
library(ggplot2)
library(vegan)
```

```{r, results='hide'}
rASV.unif <- as.data.frame(rASV)
rASV.unif = rASV.unif[,order(colSums(rASV.unif),decreasing = T)]
gunifrac.tree <- midpoint(phy_tree(psRT))
gunifracs <- GUniFrac(rASV.unif, gunifrac.tree,  alpha=c(0,0.5,1))$unifracs

d5 <- gunifracs[, , "d_0.5"]
d5.dist <- as.dist(d5)

pcoa_guniF<-cmdscale(d5, k=2, eig=T)
pcoa.var.per<-round(pcoa_guniF$eig/sum(pcoa_guniF$eig)*100, 1)
pcoa.values <- pcoa_guniF$points
pcoa.data <- data.frame(Sample=rownames(pcoa.values),
                        X=pcoa.values[,1],
                        Y=pcoa.values[,2])

pcoa.data$Location <- map$Location
pcoa.data$Supplemental_feeding <- map$Supplemental_feeding
pcoa.data$Sex <- map$Sex
pcoa.data$Lactobacillus <- map$Lactobacillus
```

Rename ASV sequences and create taxonomic vectors
```{r}
Ntaxa.df.unif <- as.data.frame(Ntaxa)
to.remove = setdiff(rownames(Ntaxa.df.unif), colnames(rASV.unif))
Ntaxa.df.unif = Ntaxa.df.unif[!row.names(Ntaxa.df.unif) %in% to.remove,]
setdiff(rownames(Ntaxa.df.unif), colnames(rASV.unif))

identical(rownames(Ntaxa.df.unif), colnames(rASV.unif))
Ntaxa.df.unif = Ntaxa.df.unif[colnames(rASV.unif),]
identical(rownames(Ntaxa.df.unif), colnames(rASV.unif))

Ntaxa.df.unif$ASV_ID <- paste("ASV", 1:nrow(Ntaxa.df.unif), sep="")
Ntaxa.df.unif$concat = paste(Ntaxa.df.unif$ASV_ID, Ntaxa.df.unif$Taxonomy, sep = "_")
rownames(Ntaxa.df.unif) = Ntaxa.df.unif$concat
names(rASV.unif) = rownames(Ntaxa.df.unif)


vec.sp<-envfit(pcoa_guniF$points , rASV.unif, perm=1000)
vec.sp.df<-
  as.data.frame(scores(vec.sp, "vectors"))
vec.sp.df$species<-rownames(vec.sp.df)
vec.sp.df.sig <- vec.sp.df[c(1:6), ]
```

```{r fig.cap = "takah\u113 gUniFrac PCoA plot by location with taxonomic vectors", fig.height = 10, fig.width = 18, dpi=300}
library(ggrepel)

GUFpcoa_vectors = ggplot(data=pcoa.data, aes(x = X, y = Y, colour = Location)) +
  geom_point(size=5, stroke=2, alpha=0.75, aes(shape=Location)) +
  xlab(paste("PCoA 1","[" ,pcoa.var.per[1], "%","]" ,sep=" ")) +
  ylab(paste("PCoA 2","[" ,pcoa.var.per[2], "%","]" ,sep=" ")) + 
  geom_vline(xintercept = 0, linetype = "dashed") + 
  geom_hline(yintercept = 0, linetype = "dashed") +
  geom_segment(data=vec.sp.df.sig, 
              aes(x=0,xend=Dim1,y=0,yend=Dim2),
               arrow = arrow(length = unit(0.3, "cm")),
               colour="red", size = 1,
               inherit.aes=FALSE) + 
 geom_text_repel(data=vec.sp.df.sig,
            aes(x=Dim1,y=Dim2,label=species),size=6.5,
            inherit.aes=FALSE, max.overlaps = Inf) +
  scale_color_manual(values = takahe_pal) + 
  scale_shape_manual(values=c(15:17,21:25)) +
  ord_theme +
  ylim(-0.50, 0.8) +  xlim(-0.44, 0.8) +
  ggtitle("ASV vectors for ordinated faecal samples")

GUFpcoa_vectors
ggsave("./ordinations/pcoa_takahe_guniF_vectors.png", width = 18, height = 10, dpi = 400, bg = "white")
```


```{r fig.cap = "takah\u113 gUniFrac PCoA plot by location", fig.height = 10, fig.width = 18, dpi=300}
GUFpcoa_location = ggplot(data=pcoa.data, aes(x = X, y = Y, colour = Location)) +
  geom_point(size=7, stroke=2, alpha=0.75, aes(shape=Location)) +
  xlab(paste("PCoA1 ","[", pcoa.var.per[1], "%", "]")) +
  ylab(paste("PCoA2 ", "[",pcoa.var.per[2], "%", "]")) + 
  geom_vline(xintercept = 0, linetype = "dashed") + 
  geom_hline(yintercept = 0, linetype = "dashed") +
  scale_color_manual(values = takahe_pal) + 
  scale_shape_manual(values=c(15:17,21:25)) +
  ord_theme +
  ggtitle("Bacterial communities by location")


GUFpcoa_location
ggsave("./ordinations/pcoa_takahe_guniF_location.png", width = 18, height = 10, bg = "white")
```

```{r fig.cap = "takah\u113 PCoA plot by lactobacillus abundance", fig.height = 10, fig.width = 18, dpi=300}
GUFpcoa_lacto = ggplot(data=pcoa.data, aes(x = X, y = Y, colour = Lactobacillus)) +
  geom_point(size=7, stroke=2, alpha=0.75, aes(shape=Location)) +
  xlab(paste("PCoA1 ","[", pcoa.var.per[1], "%", "]")) +
  ylab(paste("PCoA2 ", "[",pcoa.var.per[2], "%", "]")) + 
  geom_vline(xintercept = 0, linetype = "dashed") + 
  geom_hline(yintercept = 0, linetype = "dashed") +
  scale_color_gradient(low = "blue", high = "red", name="*Lactobacillus* abundance",
                      labels=c("Low","High"),
                       breaks=c(1000,5300)) +
  scale_shape_manual(values=c(15:17,21:25)) +
  ord_theme + theme(legend.box = "vertical") + 
  guides(shape = guide_legend(order = 1)) +
  ggtitle("Bacterial communities by *Lactobacillus* abundance")

GUFpcoa_lacto
ggsave("./ordinations/pcoa_takahe_guniF_lactobacillus.png", width = 18, height = 10, dpi=400, bg = "white")
```

```{r fig.height=30, fig.width=18}
pcoa_together = plot_grid(GUFpcoa_location, GUFpcoa_lacto, GUFpcoa_vectors, labels = c("A","B","C"), label_size = 30, ncol=1)
pcoa_together
ggsave("GUFpcoa_together.png", pcoa_together, height = 30, width = 18, dpi = 400, bg = "white")
```


#PERMANOVA

```{r}
permanova.dis <- phyloseq::distance(psR_RA, "bray")

vegan::adonis2(permanova.dis ~ Location, as(sample_data(psR_RA), "data.frame"), permutations = 9999)

vegan::adonis2(permanova.dis ~ Supplemental_feeding + Location, by = "margin", as(sample_data(psR_RA), "data.frame"), permutations = 9999)

vegan::adonis2(permanova.dis ~ Supplemental_feeding + Location, by = "term", as(sample_data(psR_RA), "data.frame"), permutations = 9999)

vegan::adonis2(permanova.dis ~ Habitat + Location, by='term', as(sample_data(psR_RA), "data.frame"), permutations = 9999)

vegan::adonis2(permanova.dis ~ Habitat + Location, by='margin', as(sample_data(psR_RA), "data.frame"), permutations = 9999)
```

```{r}
vegan::adonis2(permanova.dis ~ Supplemental_feeding, as(sample_data(psR_RA), "data.frame"), permutations = 9999)
vegan::adonis2(permanova.dis ~ Habitat, as(sample_data(psR_RA), "data.frame"), permutations = 9999)
vegan::adonis2(permanova.dis ~ Hatch_Site, as(sample_data(psR_RA), "data.frame"), permutations = 9999) #hatch site has very uneven group sizes
vegan::adonis2(permanova.dis ~ Nest_Site, as(sample_data(psR_RA), "data.frame"), permutations = 9999)
vegan::adonis2(permanova.dis ~ Sex, as(sample_data(psR_RA), "data.frame"), permutations = 9999)
vegan::adonis2(permanova.dis ~ Age, as(sample_data(psR_RA), "data.frame"), permutations = 9999)
vegan::adonis2(permanova.dis ~ Natal.Year, as(sample_data(psR_RA), "data.frame"), permutations = 9999)
vegan::adonis2(permanova.dis ~ Origin, as(sample_data(psR_RA), "data.frame"), permutations = 9999)
```

```{r}
vegan::adonis2(as.dist(d5) ~ Location, map, permutations = 9999)

vegan::adonis2(as.dist(d5) ~ Supplemental_feeding + Location, by = "margin", map, permutations = 9999)

vegan::adonis2(as.dist(d5) ~ Supplemental_feeding + Location, by = "term",map, permutations = 9999)

vegan::adonis2(as.dist(d5) ~ Supplemental_feeding, map, permutations = 9999)
vegan::adonis2(as.dist(d5) ~ Habitat, map, permutations = 9999)
vegan::adonis2(as.dist(d5) ~ Hatch_Site, map, permutations = 9999) #hatch site has very uneven group sizes
vegan::adonis2(as.dist(d5) ~ Nest_Site, map, permutations = 9999)
vegan::adonis2(as.dist(d5) ~ Sex, map, permutations = 9999)
vegan::adonis2(as.dist(d5) ~ Age, map, permutations = 9999)
vegan::adonis2(as.dist(d5) ~ Natal.Year, map, permutations = 9999)
vegan::adonis2(as.dist(d5) ~ Origin, map, permutations = 9999)
```

Pairwise PERMANOVA with Benjamini-Hochberg adjustment

```{r results=F, message=F, warning=F}
library(pairwiseAdonis); packageVersion("pairwiseAdonis")
library(permute)
```


```{r}
pairwiseAdonis::pairwise.adonis(permanova.dis, factors = map$Location, p.adjust.m = 'BH')
pairwiseAdonis::pairwise.adonis(as.dist(d5), factors = map$Location, p.adjust.m = 'BH')
```

Check that any significance detected in PERMANOVA is not due to uneven dispersion between groups.

```{r fig.height = 10, fig.width = 15, dpi=300}
beta <- vegan::betadisper(permanova.dis, map$Location)
beta
vegan::permutest(beta, permutations = how(nperm = 9999))

beta2 <- vegan::betadisper(as.dist(d5), map$Location)
beta2
vegan::permutest(beta2, permutations = how(nperm = 9999))
```

Exploring supplemental feeding and habitat type

```{r fig.height = 10, fig.width = 15, dpi=300}
pairwiseAdonis::pairwise.adonis(permanova.dis, factors = map$Supplemental_feeding, p.adjust.m = 'BH')
pairwiseAdonis::pairwise.adonis(as.dist(d5), factors = map$Supplemental_feeding, p.adjust.m = 'BH')
```

Remove smaller 'occasional' group

```{r}
new_psR_SF = subset_samples(psR, SampleID != "Moehau" & SampleID != "Puti" & SampleID != "Tamahine" & SampleID != "Tautari")
metadata_SF <- as(sample_data(new_psR_SF), "data.frame")
permanova.dis_SF <- phyloseq::distance(new_psR_SF, "bray")
vegan::adonis2(permanova.dis_SF ~ Supplemental_feeding, as(sample_data(new_psR_SF), "data.frame"), permutations = 9999)


psRT_SF = subset_samples(psRT, SampleID != "Moehau" & SampleID != "Puti" & SampleID != "Tamahine" & SampleID != "Tautari")
rASV.unif.SF <- as.data.frame(otu_table(psRT_SF))
rASV.unif.SF = rASV.unif.SF[,order(colSums(rASV.unif.SF),decreasing = T)]
gunifrac.tree.SF <- midpoint(phy_tree(psRT_SF))
gunifracs.SF <- GUniFrac(rASV.unif.SF, gunifrac.tree.SF,  alpha=c(0,0.5,1))$unifracs

d5.SF <- gunifracs.SF[, , "d_0.5"]
d5.dist.SF <- as.dist(d5.SF)
map.sf <- data.frame(sample_data(psRT_SF))
vegan::adonis2(d5.dist.SF ~ Supplemental_feeding, map.sf, permutations = 9999)

```

```{r}
betaSF <- vegan::betadisper(permanova.dis, map$Supplemental_feeding)
betaSF
vegan::permutest(betaSF, permutations = how(nperm = 9999))

betaSF <- vegan::betadisper(permanova.dis_SF, metadata_SF$Supplemental_feeding)
betaSF
vegan::permutest(betaSF, permutations = how(nperm = 9999))

betaSF2 <- vegan::betadisper(as.dist(d5), map$Supplemental_feeding)
betaSF2
vegan::permutest(betaSF2, permutations = how(nperm = 9999))

betaSF2 <- vegan::betadisper(d5.dist.SF, map.sf$Supplemental_feeding)
betaSF2
vegan::permutest(betaSF2, permutations = how(nperm = 9999))
```

```{r}
betaH <- vegan::betadisper(permanova.dis, map$Habitat)
betaH
vegan::permutest(betaH, permutations = how(nperm = 9999))

betaH2 <- vegan::betadisper(as.dist(d5), map$Habitat)
betaH2
vegan::permutest(betaH2, permutations = how(nperm = 9999))
```


#Compare core genera relative sequence abundance between locations

```{r results=F, message=F, warning=F}
library(ggplot2)
library(ape); packageVersion("ape")
library(microbiome)
library(dplyr)
library(vegan); packageVersion("vegan")
library(phyloseq)
library(reshape2); packageVersion("reshape2")
library(ggpubr)
```

```{r}
ps.bxa <- psR
taxic <- as.data.frame(ps.bxa@tax_table) 
taxic$OTU <- row.names(taxic)
taxmat <- as.matrix(taxic)
# convert into phyloseq compaitble file.
new.tax <- tax_table(taxmat)  
tax_table(ps.bxa) <- new.tax 
pseq.fam.bx <- microbiome::aggregate_rare(ps.bxa,level = "Genus",detection = 0.01/100,prevalence = 75/100) 

metadf1 <- map
# create comaprision
lev1 <- levels(as.factor(metadf1$Location)) # identify the grouping variable you want to compare between
# make a pairwise list that we want to compare.
L.pairs1 <- combn(seq_along(lev1), 2, simplify = FALSE, FUN = function(i)lev1[i])
pseq.famrel.bx <- microbiome::transform(pseq.fam.bx, "compositional") #compositional is relative abundace
ps.bx.m <- psmelt(pseq.famrel.bx)
ps.bx.m.sub <- subset(ps.bx.m, OTU != "Other" & OTU != "Unclassified")
```

```{r}
compare.mean <- compare_means(formula = Abundance ~ Location,group.by = "OTU", data=ps.bx.m.sub,p.adjust.method = "BH", method = "kruskal.test")
compare.mean 

write.csv(compare.mean, "./core/compare.means.csv")

genus.p.df <- compare_means(formula = Abundance ~ Location,group.by = "OTU", data=ps.bx.m, method = "kruskal.test")
genus.p.df
```

```{r results=F, message=F, warning=F}
library(hrbrthemes)
library(extrafont); packageVersion("extrafont")
```
     
                                                                                
```{r fig.cap = "Differential abundance of core genera between locations",fig.width=15, fig.height=15}
labels = c(paste('Bacteroides\np = 0.01*'),
           paste("Bifidobacterium\np < 0.001***"),
           paste("Campylobacter\np = 0.23"),
           paste("Clostridium sensu stricto 1\np = 0.02*"),
           paste("Lactobacillus\np = 0.002**"),
           paste("Prevotella\np = 0.002**"),
           paste("Pseudomonas\np = 0.13"),
           paste("Solobacterium\np = 0.004**"))

ps.bx.m.sub$OTU <- factor(ps.bx.m.sub$OTU, levels = c("Bacteroides", "Bifidobacterium", "Campylobacter","Clostridium sensu stricto 1", "Lactobacillus", "Prevotella", "Pseudomonas","Solobacterium"), 
                          labels = labels)

ps.bx.m.sub$Location = as.factor(ps.bx.m.sub$Location)
  
px2 <- ggboxplot(ps.bx.m.sub, x = "Location", y = "Abundance", 
                 add = "dotplot", 
                 fill = "Location", 
                 palette = c("#DD3C51", "#313657", "#51806a","#1F6683", "#83A552","#6C90B9","#E2939E", "#D1C7B5"), 
                 ylab = "Relative sequence abundance") + scale_y_log10() + 
  theme_ipsum() + theme(axis.text.x = element_blank(),

          axis.ticks.x = element_blank(),

          axis.title.y=element_text(size=23),
          
          axis.title.x = element_text(size=23),
          
          strip.text.x = element_text(size = 23),

          axis.ticks.y = element_line(color="black",size=0.35,linetype=1),

          axis.line.x=element_line(color="black",size=0.15,linetype=1),

          axis.line.y=element_line(color="black",size=0.15,linetype=1),

          panel.grid.major = element_blank(),

          panel.grid.minor = element_blank(),

          panel.background = element_blank(),

          legend.box.background = element_rect(),

          legend.box.margin = margin(5, 5, 5, 5),

          legend.position="bottom",

          legend.title = element_text(face = "bold", size = 25),
          
          legend.text = element_text(size = 23)) +
  
   facet_wrap(~OTU, scales = "free") 

  
px2
ggsave("./core/genus_abundances_location.png", width = 15, height = 15, dpi = 400)
```

#ASV differential abundance testing

```{r results=F, message=F, warning=F}
library(plyr); packageVersion("plyr")
library(ggplot2); packageVersion("ggplot2")
library(dplyr); packageVersion("dplyr")
library(microbiome); packageVersion("microbiome")
library(phyloseq); packageVersion("phyloseq")
library(DESeq2); packageVersion("DESeq2")
library(ggpubr); packageVersion("ggpubr")
library(hrbrthemes); packageVersion("hrbrthemes")
```

```{r}
my_theme <- theme_ipsum() + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), axis.text.y = element_text(face="italic"))

ps2.ASV.df = as.data.frame(otu_table(ps2))
ps2.ASV.df = ps2.ASV.df[,order(colSums(ps2.ASV.df),decreasing = T)]
ps2.taxa.df = as.data.frame(tax_table(ps2))

identical(rownames(ps2.taxa.df), colnames(ps2.ASV.df))
setdiff(rownames(ps2.taxa.df), colnames(ps2.ASV.df))
ps2.taxa.df = ps2.taxa.df[colnames(ps2.ASV.df),]

ps2.taxa.df$ASV_ID <- paste("ASV_", 1:nrow(ps2.taxa.df), sep="")
ps2.taxa.df$concat = paste(ps2.taxa.df$ASV_ID, ps2.taxa.df$Taxonomy, sep = "_")
rownames(ps2.taxa.df) = ps2.taxa.df$concat
names(ps2.ASV.df) = rownames(ps2.taxa.df)

ps2.taxa.df$ASV_ID = NULL
ps2.taxa.df$concat = NULL
ps2.taxa.df = as.matrix(ps2.taxa.df)

ps2_edit <- phyloseq(otu_table(ps2.ASV.df, taxa_are_rows = F),
                     tax_table(ps2.taxa.df),
                     sample_data(map))
ps2_edit
```

```{r}
ps.sta <- ps2_edit
sample_data(ps.sta)$Location <- revalue(sample_data(ps.sta)$Location, c("Burwood Centre" = "Burwood.Centre", "Cape Sanctuary" = "Cape.Sanctuary", "Foveaux Strait Is" = "Foveaux.Strait.Is", "Mana Island" = "Mana.Island", "Motutapu Island" = "Motutapu.Island", "Murchison Mountain" = "Murchison.Mountain", "Rotoroa Island" = "Rotoroa.Island", "Tiritiri Matangi Is" = "Tiritiri.Matangi.Is"))

ps.st1.bc.mm.0 = subset_samples(ps.sta, Location != "Cape.Sanctuary" & Location != "Foveaux.Strait.Is" & Location != "Mana.Island" & Location != "Motutapu.Island" & Location != "Rotoroa.Island" & Location != "Tiritiri.Matangi.Is")
head(sample_data(ps.st1.bc.mm.0)$Location, 10)
ps.st1.bc.mm.0.1 = filter_taxa(ps.st1.bc.mm.0, function(x) sum(x > 3) > (0.05*length(x)), TRUE)
ps.st1.bc.mm <- ps.st1.bc.mm.0.1
```

```{r}
meta.st <- meta(ps.st1.bc.mm.0)
meta.st$Location <- as.factor(meta.st$Location)
diagdds_st.bc.mm = phyloseq_to_deseq2(ps.st1.bc.mm, ~ Location)
 gm_mean = function(x, na.rm=TRUE){
     exp(sum(log(x[x > 0]), na.rm=na.rm) / length(x))
 }
 geoMeans = apply(counts(diagdds_st.bc.mm), 1, gm_mean)
 diagdds_st.bc.mm = estimateSizeFactors(diagdds_st.bc.mm, geoMeans = geoMeans)
 dds_st.bc.mm = DESeq(diagdds_st.bc.mm, test="Wald", fitType="local")
```

```{r}
otu.ab1 <- abundances(ps.st1.bc.mm)

res1 = results(dds_st.bc.mm, cooksCutoff = FALSE)
res_tax1 = cbind(as.data.frame(res1), as.matrix(rownames(otu.ab1)), OTU = rownames(res1))
res_tax1 = cbind(as(res_tax1, "data.frame"), as(tax_table(ps.st1.bc.mm)[rownames(res_tax1), ], "matrix"))
res_tax_sig1 = subset(res_tax1, padj < 0.01 & 0 < abs(log2FoldChange))
res_tax1$Significant <- ifelse(rownames(res_tax1) %in% rownames(res_tax_sig1) , "Yes", "No")
res_tax1$Significant[is.na(res_tax1$Significant)] <- "No"
sig_res1 <- res_tax1[rownames(res_tax_sig1),"OTU"]
res_table1 <- data.frame(res_tax_sig1$baseMean , res_tax_sig1$log2FoldChange,res_tax_sig1$padj)
row.names(res_table1) <- rownames(res_tax_sig1)

data_to_write1 <-res_tax_sig1[,c("baseMean","log2FoldChange","pvalue","padj","Phylum", "Class", "Order", "Family", "Genus","Species","Taxonomy","OTU")]
data_to_write1$DifferentiallyAbundant <-levels(meta.st[,"Location"])[as.numeric(data_to_write1$log2FoldChange>0)+1]

# Total numer of OTUs DA
nrow(data_to_write1) #differentially abundant ASVs
length(unique(data_to_write1$Genus)) #how many genera
length(unique(data_to_write1$Taxonomy)) #how many species
write.csv(data_to_write1,"./tables/deseq_comparison_mm_vs_bc.csv",sep="")
df1 <- mutate(data_to_write1, Taxonomy, Taxonomy= paste(data_to_write1$Taxonomy ))
```

```{r fig.cap = "Differential abundance of significant ASVs between Murchison Mountain and Burwood Centre", fig.width=12, fig.height=13, dpi=300}
bc.mm.st <- ggplot(df1, aes(log2FoldChange, Taxonomy)) + 
  geom_point(aes(color = DifferentiallyAbundant), shape = 21, size = 3, stroke = 2) + 
  scale_color_manual(values= c("#DD3C51","#1F6683"), name = "Location", labels = c("Burwood Centre", "Murchison Mountain")) + my_theme + 
  labs(y = "Species") +
  theme(axis.text.y = element_text(face="italic", size = 15),
        panel.border = element_rect(color="black",size=0.7,linetype=1, fill = "NA"), 
        panel.background = element_blank(),
        axis.title.y = element_text(size = 35),
        axis.title.x = element_text(size=23),
        legend.box.background = element_rect(),
        legend.box.margin = margin(5, 5, 5, 5),
        legend.title = element_text(face = "bold", size = 23),
        legend.text = element_text(size = 20), 
        legend.position = "top") +  geom_vline(xintercept = 0) 
bc.mm.st 
ggsave("./core/Deseq_bc_mm.png", width = 12, height = 13, dpi=400)
```


```{r}
save.image(file = "takahe_data.Rdata")
```