microbiome_diet_trends.Rmd

---
title: "Weekly Microbiome Composition and Dietary Intake"
output:
  workflowr::wflow_html:
    toc: true
editor_options:
  chunk_output_type: console
---

```{r data, include=FALSE, echo=FALSE}

# load packages
source("code/load_packages.R")

# get data
source("code/get_cleaned_data.R")

theme_set(theme_bw())
pal = "Set1"
scale_colour_discrete <-  function(palname=pal, ...){
  scale_colour_brewer(palette=palname, ...)
}
scale_fill_discrete <-  function(palname=pal, ...){
  scale_fill_brewer(palette=palname, ...)
}

knitr::opts_chunk$set(out.width = "200%")

```

# Microbiome Composition

```{r}

mphyseq = psmelt(phylo_data)
mphyseq2 <- mphyseq %>%
  dplyr::group_by(SubjectID, Week) %>%
  dplyr::mutate(Total = sum(Abundance)) %>%
  dplyr::ungroup()%>%
  dplyr::group_by(SubjectID, Week, Phylum) %>%
  dplyr::mutate(PhylumAbund = sum(Abundance),
                RelAbund = PhylumAbund/Total)
  
mphyseq2 <- mphyseq2 %>% distinct(SubjectID, Week, Phylum, .keep_all = T)

keepVar <- c("SubjectID", "Week", "Phylum", "Abundance", "RelAbund")

mphyseq2 <- mphyseq2[, keepVar]

# take out "__" at start of names
mphyseq2$Phylum <- substring(mphyseq2$Phylum, 3)

# Create New Other category for plotting
mphyseq3 <- mphyseq2 %>% 
  dplyr::group_by(SubjectID, Week, Phylum) %>%
  dplyr::summarise(RelAbund = sum(RelAbund))
other <- mphyseq3 %>% 
  dplyr::group_by(SubjectID, Week) %>% 
  dplyr::summarise(RelAbund = 1 - sum(RelAbund))

other$Phylum <- "Other"
other <- other %>% select(SubjectID, Week,  Phylum, RelAbund)

mphyseq2 <- full_join(mphyseq3, other)


# sort by highest average relative abundance
ph <- mphyseq2 %>%
  dplyr::group_by(Phylum) %>%
  dplyr::summarize(M = mean(RelAbund, na.rm=T))

micro_ord <- ph$Phylum[order(ph$M, decreasing = F)]

mphyseq2$Phylum <- factor(mphyseq2$Phylum, levels = rev(micro_ord))

# fix missing data and fill-out
MIS <- mphyseq2 %>%
  group_by(SubjectID)%>%tidyr::expand(Week, Phylum)

micro_data <- full_join(mphyseq2, MIS)

# add week 1 missing as 0
micro_data$RelAbund[micro_data$Week==1][is.na(micro_data$RelAbund[micro_data$Week==1])] <- 0

micro_data <- micro_data%>%
  group_by(SubjectID, Phylum)%>%
  fill(RelAbund)

micro_data$Week <- as.numeric(micro_data$Week)

# create order of subjects
so <- distinct(microbiome_data$meta.dat, SubjectID, .keep_all = T)

subjectorder <- so$SubjectID[order(so$Intervention, decreasing = F)]

micro_data$SubjectID <- factor(micro_data$SubjectID,
                             levels = subjectorder,
                             labels=c(1:11))


# get right number of colors for plotting
no_cols <- length(unique(micro_data$Phylum))

## Some Colors
colors_micro <- rev(c("grey90", rev(c("#00a2f2",  "#c91acb", "#7f5940", "#cc5200", "#00d957", "#40202d", "#e60099", "#006fa6", "#f29d3d", "#300059"))))

# Intervention ID variable
ids <- 1:6
micro_data$Intervention <- ifelse(micro_data$SubjectID %in% ids, "Group A", "Group B")


# make the plot
micro_plot<-ggplot(data = micro_data, aes(x=Week, y = RelAbund, fill=Phylum)) +
  geom_area(stat = "identity") +
  facet_grid(.~Intervention + SubjectID, scales = "free") +
  scale_fill_manual(values = colors_micro) +
  lims(x=c(0.99, 4.01))+
  theme_classic() +
  theme(strip.text.x = element_text(angle = 0, size = 11, face = "italic"),
        axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        axis.text.y = element_text(size = 10),
        axis.title = element_text(size = 10),
        plot.title = element_text(hjust = 0.5),
        axis.title.x = element_blank(),
        strip.background = element_blank(),
        legend.position = "right",
        legend.text = element_text(size = 7),
        legend.title = element_blank(),
        panel.spacing.x=unit(0.001, "lines")) +
  guides(fill = guide_legend(reverse = F,
                             keywidth = 0.5,
                             keyheight = 0.5,
                             ncol = 1)) +
  labs(y="Relative Abundance",
       title="Gut Microbiome, Phylum Level",
       tag="A")
micro_plot

# #Next, change strip color by intervention group
# g <- ggplot_gtable(ggplot_build(micro_plot))
# strip_both <- which(grepl('strip-', g$layout$name))
# fills <- c(rep("white", 6), rep("grey80", 5))
# k <- 1
# for (i in strip_both) {
#   j <- which(grepl('rect', g$grobs[[i]]$grobs[[1]]$childrenOrder))
#   g$grobs[[i]]$grobs[[1]]$children[[j]]$gp$fill <- fills[k]
#   k <- k+1
# }
# grid::grid.draw(g)
# 
# 
# tag_facet2 <-  function(p, open=" ", close = " ",
#          tag_pool = letters,
#          x = 0, y = 0.5,
#          hjust = 0, vjust = 0.5, 
#          fontface = 2, ...){
#   
#   gb <- ggplot_build(p)
#   lay <- gb$layout$layout
#   nm <- names(gb$layout$facet$params$rows)
#   
#   tags <- paste0(open,tag_pool[unique(lay$COL)],close)
#   
#   tl <- lapply(tags, grid::textGrob, x=x, y=y,
#                hjust=hjust, vjust=vjust, gp=grid::gpar(fontface=fontface))
#   
#   g <- ggplot_gtable(gb)
#   g <- gtable::gtable_add_rows(g, grid::unit(1,"line"), pos = 0)
#   lm <- unique(g$layout[grepl("panel",g$layout$name), "l"])
#   g <- gtable::gtable_add_grob(g, grobs = tl, t=1, l=lm)
#   grid::grid.newpage()
#   grid::grid.draw(g)
# }
# 
# IntGrp <- c(rep("A", 6), rep("B", 5))
# micro_plot2<-micro_plot + theme(legend.position = "none")
# micro_plot2
# tag_facet2(micro_plot2, tag_pool = IntGrp)


  
```

Note: Subjects are ordered by Intervention:
1 to 6 are in Group A
7-11 are in group B

# Make Dietary Figures

```{r include=F}


diet.data <- read_xlsx("data/analysis-data/Dietary_Variables.xlsx")
diet.data <- diet.data %>%
  mutate(SubjectID = as.numeric(substr(SubjectID, 7,11)),
         Week = RecallNo)


```

## Food Groups

```{r}

food_var <- colnames(diet.data)[4:12]

food_data <- as_tibble(diet.data[, c("SubjectID", "Week", food_var)])


# need to fill in "missing" data
MIS <- tidyr::expand(food_data, SubjectID, Week)

food_data <- full_join(food_data, MIS)

food_data <- food_data%>%
  group_by(SubjectID)%>%
  fill(`Fats Oils and Salad Dressings`:`Grain Product`)


id <- paste0("id.", food_data$SubjectID,".wk.",food_data$Week)

food_data$MISSING <- apply(food_data, 1,
                           FUN = function(x){ sum(is.na(x)) })

food_data <- data.frame(t(food_data[,-c(1:2)]))
colnames(food_data) <- id
rownames(food_data) <- c(food_var, "MISSING")

food_data <- apply(food_data, c(1,2), FUN=function(x){ifelse(is.na(x), 0, x)})
food_data <- data.frame(food_data)
# Compute relative abundance
food_data <- sweep(food_data, 2, colSums(food_data), "/")
food_data[is.na(food_data)] <- 0

# sort by highest average relative abundance
food_data <- food_data[order(rowMeans(food_data), decreasing = F),]

# make food ordering factor
food_ord_factor <- as.character(rownames(food_data))
food_ord_factor <- food_ord_factor[food_ord_factor != "MISSING"]
food_ord_factor <- c("MISSING",food_ord_factor)


plot3 <- as.data.frame(t(food_data))
plot3 <- rownames_to_column(plot3, var = "SampleID")
plot3 <- reshape2::melt(plot3, id = "SampleID", variable.name = "Food")

# combine all "<x% abundance" foods into one for plotting
#plot3 <- plot3 %>% group_by(SampleID, Food) %>% dplyr::summarise(newvalue = sum(value))

# Extract ids and weeks
#plot3$SubjectID <- str_sub(plot3$SampleID, 4,7)
so <- distinct(microbiome_data$meta.dat, SubjectID, .keep_all = T)

subjectorder <- so$SubjectID[order(so$Intervention, decreasing = F)]

plot3$SubjectID <- factor(str_sub(plot3$SampleID, 4,7),
                          levels = subjectorder)

plot3$Week <- as.numeric(str_sub(plot3$SampleID, 12, 13))
# recode FOOD
plot3$Food <- as.factor(plot3$Food)
plot3$Food <- factor(plot3$Food, levels = rev(food_ord_factor))
# set seed to get nice colors
set.seed(3)

# get right number of colors for plotting
no_cols <- length(unique(plot3$Food))

## Some Colors
colors_food <- c("#00a2f2",  "#c91acb", "#7f5940", "#cc5200", "#00d957", "#40202d", "#e60099", "#006fa6", "#f29d3d", "grey90")



# make the plot
food_plot<-ggplot(data = plot3, aes(x=Week, y = value, fill=Food)) +
  geom_area(stat = "identity") +
  facet_grid(.~SubjectID, scales = "free") +
  scale_fill_manual(values = colors_food) +
  lims(x=c(0.99, 4.01))+
  theme_classic() +
  theme(strip.text.x = element_blank(),
        axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        axis.text.y = element_text(size = 10),
        axis.title = element_text(size = 10),
        plot.title = element_text(hjust = 0.5),
        axis.title.x = element_blank(),
        strip.text = element_blank(),
        strip.background = element_blank(),
        legend.position = "right",
        legend.text = element_text(size = 7),
        legend.title = element_blank(),
        panel.spacing.x=unit(0.001, "lines")) +
  guides(fill = guide_legend(reverse = F,
                             keywidth = 0.5,
                             keyheight = 0.5,
                             ncol = 1)) +
  #nrow = 5)) + # for full page figure
  #nrow = 1)) + # for slide figure
  labs(y="Relative Abundance",
       title="Dietary Food Groups",
       tag="B")
food_plot


```


## Nutrients

```{r}

# Dietary Nurtrients


food_var <- colnames(diet.data)[14:54]


food_data <- as_tibble(diet.data[, c("SubjectID", "Week", food_var)])
# need to fill in "missing" data
MIS <- tidyr::expand(food_data, SubjectID, Week)

food_data <- full_join(food_data, MIS)

food_data <- food_data%>%
  group_by(SubjectID)%>%
  fill(`Carbohydrates`:`Added Vitamin B-12`)

id <- paste0("id.", food_data$SubjectID,".wk.",food_data$Week)

food_data$MISSING <- apply(food_data, 1,
                           FUN = function(x){ sum(is.na(x)) })

food_data <- data.frame(t(food_data[,-c(1:2)]))
colnames(food_data) <- id
rownames(food_data) <- c(food_var, "MISSING")

food_data <- apply(food_data, c(1,2), FUN=function(x){ifelse(is.na(x), 0, x)})
food_data <- data.frame(food_data)
# Compute relative abundance
food_data <- sweep(food_data, 2, colSums(food_data), '/')

# sort by highest average relative abundance
food_data <- food_data[order(rowMeans(food_data), decreasing = F),]
rn <- rownames(food_data)
#food_data <- food_data[ rn[c(1:8, 10, 9)], ]

# make food ordering factor
food_ord_factor <- as.character(rownames(food_data))
food_ord_factor <- food_ord_factor[food_ord_factor != "MISSING"]
food_ord_factor <- c("MISSING",food_ord_factor)

plot3 <- as.data.frame(t(food_data))
plot3 <- rownames_to_column(plot3, var = "SampleID")
plot3 <- reshape2::melt(plot3, id = "SampleID", variable.name = "Food")

# Extract ids and weeks

#plot3$SubjectID <- str_sub(plot3$SampleID, 4,7)
so <- distinct(microbiome_data$meta.dat, SubjectID, .keep_all = T)

subjectorder <- so$SubjectID[order(so$Intervention, decreasing = F)]

plot3$SubjectID <- factor(str_sub(plot3$SampleID, 4,7),
                          levels = subjectorder)
plot3$Week <- as.numeric(str_sub(plot3$SampleID, 12, 13))
# recode FOOD
plot3$Food <- as.factor(plot3$Food)
plot3$Food <- factor(plot3$Food, levels = rev(food_ord_factor))
# set seed to get nice colors
set.seed(3)

# get right number of colors for plotting
no_cols <- length(unique(plot3$Food))

## Some Colors
colors_food <- c("#00a2f2",  "#c91acb", "#7f5940", "#cc5200", "#00d957", "#40202d", "#e60099", "#006fa6", "#f29d3d", "#300059", "#566573", "#336655", "#83008c", "#d9a3aa", "#400009", "#0020f2", "#a3d936", "#8091ff", "#fbffbf", "#00ffcc", "#8c4f46", "#354020", "#39c3e6", "#333a66", "#ff0000", "#6a8040", "#a6538a", "#402910", "#730f00", "#0a4d00", "#ffe1bf", "#a3d9b1", "#003033", "#f29979", "#00b3a7", "#cbace6", "#bfd9ff", "#bf0000", "#293aa6", "#594943", "#e5c339", "grey90")



# make the plot
nutr_plot<-ggplot(data = plot3, aes(x=Week, y = value, fill=Food)) +
  geom_area(stat = "identity") +
  facet_grid(.~SubjectID, scales = "free") +
  scale_fill_manual(values = colors_food) +
  lims(x=c(0.99, 4.01))+
  theme_classic() +
  theme(strip.text.x = element_blank(),
        axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        axis.text.y = element_text(size = 10),
        axis.title = element_text(size = 10),
        plot.title = element_text(hjust = 0.5),
        axis.title.x = element_blank(),
        strip.background = element_blank(),
        legend.position = "right",
        legend.text = element_text(size = 7),
        legend.title = element_blank(),
        panel.spacing.x=unit(0.01, "lines")) +
  guides(fill = guide_legend(reverse = F,
                             keywidth = 0.5,
                             keyheight = 0.5,
                             ncol = 1)) +
  #nrow = 5)) + # for full page figure
  #nrow = 1)) + # for slide figure
  labs(y="Relative Abundance",
       title="Dietary Macronutrients and Micronutrients",
       tag="C")
nutr_plot



```


## Saving plot

```{r}

##### MAKE THE ACTUAL FIGURE ###########
# combine into one big plot
get_legend <- function(p) {
  tmp <- ggplot_gtable(ggplot_build(p))
  leg <- which(sapply(tmp$grobs, function(x) x$name) == "guide-box")
  legend <- tmp$grobs[[leg]]
  legend
}

micro_plot_leg <- get_legend(micro_plot)
food_plot_leg <- get_legend(food_plot)
nutr_plot_leg <- get_legend(nutr_plot)

# and replot suppressing the legend
micro_plot_1 <- micro_plot + theme(legend.position='none')
food_plot_1 <- food_plot + theme(legend.position='none')
nutr_plot_1 <- nutr_plot + theme(legend.position='none')

p <- micro_plot_1 + food_plot_1 + nutr_plot_1 + plot_layout(ncol=1)

p
ggsave("fig/figure4.pdf", p, units="in", width=7.9,height=6.5)



library(cowplot)
bigplotlegend <- plot_grid(micro_plot_leg, food_plot_leg, nutr_plot_leg, nrow =1, align = "h")

save_plot("fig/figure4_legend.pdf", bigplotlegend, base_width = 13.25, base_height = 5)


```