alpha_div_and_pb_ratio.rmd

---
title: "Alpha diversity and P/B ratio"
author: "Lan Huong Nguyen"
date: "11/19/2018"
output: 
  BiocStyle::html_document:
    toc_float: true
    df_print: paged
---

```{r setup, include=FALSE}
#rm(list = ls())
knitr::opts_chunk$set(
  echo = TRUE, warning = FALSE, message = FALSE, 
  fig.path = "../figs/alphadiv_pbra/", dev=c('png', 'pdf'),
  fig.align = 'center', fig.wide = TRUE, 
  fig.width = 8, fig.height = 6
)
```

# Set up

```{r}
.packages <- c("gridExtra","RColorBrewer", "viridis", "phyloseq", 
               "tidyverse", "ggsignif")
sapply(.packages, require, character.only = TRUE)

options(stringsAsFactors = FALSE)
theme_set(theme_bw())
theme_update(
  text = element_text(size = 25),
  strip.background = element_rect(fill = "grey95"))
```



## Load data

```{r}
(ps0 <- readRDS("../data/phyloseq/diet_participants_with_adherence.rds"))

# calculate adherence, and Subj_Group variables
newSam_data <- sample_data(ps0) %>%
  data.frame() %>%
  mutate_at(vars(starts_with("adherence")), as.numeric) %>%
  mutate(adherence.raw = rowMeans(select(., adherence_3_months, adherence_6_months, adherence_12_months), na.rm = T),
         adherence.pct = 100-adherence.raw, 
         adherence.changeFromBL = adherence_baseline - adherence.raw,
         Subj_Group = factor(paste0(SubjectID, "_", Timepoint)), 
         rowname = MeasID) %>%
  column_to_rownames("rowname")


sample_data(ps0) <- newSam_data

```


## Filter out samples with insufficient reads or no weight-loss data at 12 months

```{r}
minSampleDepth <- 10000


(ps0 <- subset_samples(ps0, sample_sums(ps0) > minSampleDepth))
(ps0 <- subset_samples(ps0, !(SubjectID == "1139" & Timepoint == "10 Week")))
(ps0 <- subset_samples(ps0, !is.na(Cat.12Mo)))
ps0 <- subset_taxa(ps0, taxa_sums(ps0) > 0)

```

## Twice-sequenced samples

Find biological samples which were sequenced twice:

```{r}
(sample_data(ps0) %>% 
   group_by(SampleID) %>% 
   dplyr::summarise(
     count = n(), 
     meas = paste0(MeasID, collapse = "; ")
     ) %>%
   filter(count > 1))

```


Combine repeated samples by computing the mean 

```{r}
collapse_samples <- function(physeq, sample_id = "SampleID", fun = mean){
  sampledata_collapsed <- sample_data(physeq) %>%
    dplyr::select(-Lane, -MeasID, -OrigSampleDepth) %>%
    distinct() 
  rownames(sampledata_collapsed) <- sampledata_collapsed[[sample_id]]
  seqtab <- data.frame(as(otu_table(physeq), "matrix"))
  if(!taxa_are_rows(physeq)) seqtab <- t(seqtab)
  
  sample_ids <- physeq@sam_data[[sample_id]]
  seqtab_collapsed <- sapply(unique(sample_ids), function(id){ 
    if(sum(sample_ids == id) == 1) return(seqtab[, sample_ids == id])
    apply(seqtab[, sample_ids == id], 1, fun)
  })
  rownames(seqtab_collapsed) <- rownames(seqtab)
  physeq_collapsed <- phyloseq(
    otu_table(seqtab_collapsed, taxa_are_rows = TRUE),
    sample_data(sampledata_collapsed), 
    tax_table(physeq), phy_tree(physeq))
  return(physeq_collapsed)
}
```



```{r}
ps0 <- collapse_samples(ps0)
ps0
sample_data(ps0)$Cat.12Mo <- 
  factor(sample_data(ps0)$Cat.12Mo, levels = c("US", "MS", "VS"))
```

# Separate into different phyloseq objects for each cohort
```{r}
(ps2 <- subset_samples(ps0, Cohort == 2))
(ps2 <- subset_taxa(ps2, taxa_sums(ps2) > 0))

(ps3 <- subset_samples(ps0, Cohort == 3))
(ps3 <- subset_taxa(ps3, taxa_sums(ps3) > 0))

```

Get only baseline samples

```{r}
(ps.bl <- subset_samples(ps0, Timepoint == "Baseline"))
```

```{r}
table(sample_data(ps.bl)$Subject)

length(unique(sample_data(ps.bl)$Subject))

```


```{r}
subj_with_mult_days <- sample_data(ps.bl) %>% 
  group_by(SubjectID) %>%
  dplyr::summarise(
    no_uniq_days = length(unique(SampleDate)),
    uniq_days = paste0(unique(as.character(sort(as.Date(SampleDate)))),
                       collapse = "; ")
  ) %>%
  filter(no_uniq_days > 1)
head(subj_with_mult_days)



subj_with_mult_days_cohort3 <- sample_data(ps3) %>% 
  group_by(SubjectID) %>%
  dplyr::summarise(
    no_uniq_days = length(unique(SampleDate)),
    uniq_days = paste0(unique(as.character(sort(as.Date(SampleDate)))),
                       collapse = "; ")
  ) %>%
  filter(no_uniq_days > 1)
head(subj_with_mult_days_cohort3)
```

```{r}
ps.bl <- ps.bl %>% subset_samples(SubjectID %in% subj_with_mult_days$SubjectID)
ps.bl

ps3.bl <- ps3 %>% subset_samples(SubjectID %in% subj_with_mult_days_cohort3$SubjectID)
ps3.bl
```

```{r}
diet_cols <- c(brewer.pal(9, "Set1")[2], brewer.pal(9, "Set1")[5])
```


# Provotella / Bacteroides ratio

Calculate PB ratio per sample and then take average for each subject
```{r}
taxtab <- data.frame(ps0@tax_table) %>%
  rownames_to_column("SeqName") %>%
  dplyr::select(-Seq)

prevotella_bacteroides <- taxtab %>%
  filter( grepl("Prevo", Genus) | grepl("Bacter", Genus)) %>%
  filter(!grepl("Prevotellaceae", Genus))

table(prevotella_bacteroides$Genus)
```


```{r}
#find all taxa identified as Prevotella or Bacteroides
prevotella_bacteroides <- prevotella_bacteroides %>%
  mutate(Genus = ifelse(grepl("Prevo", Genus) , "Prevotella", Genus)) 
table(prevotella_bacteroides$Genus)
```

```{r}
#get phyloseq object with only P and B
ps_PB <- subset_taxa(
  ps0, taxa_names(ps0) %in% prevotella_bacteroides$SeqName)
ps_PB
```


```{r}
tax_table(ps_PB)[, "Genus"] <- 
  sapply(tax_table(ps_PB)[, "Genus"], function(x) {
    ifelse(grepl("Prevo", x), "Prevotella", x)
  })
table(tax_table(ps_PB)[, "Genus"])
```


```{r}
ps_PB <- merge_taxa(ps_PB, eqtaxa = "Genus")
ps_PB
```

## Calculate P/B ratio

```{r}
pb_ratio <- data.frame(
  (as(otu_table(ps_PB), "matrix")), check.names = FALSE) %>%
  rownames_to_column("SeqName") %>%
  left_join(
    data.frame(tax_table(ps_PB)) %>%
      rownames_to_column("SeqName") %>%
      dplyr::select(SeqName, Genus)) %>%
  group_by(Genus) %>%
  column_to_rownames("SeqName") %>%
  summarise_all(.funs = sum) %>%
  column_to_rownames("Genus")

pb_ratio <- t(pb_ratio) %>%
  as.data.frame() %>%
  rownames_to_column("SampleID") %>%
  mutate(
    Ratio = ifelse(Prevotella == 0 & Bacteroides == 0, 0, Prevotella/Bacteroides),
    logRatio = log10((Prevotella + 1)/(Bacteroides+1))) %>%
  left_join(data.frame(sample_data(ps_PB)) %>%
              dplyr::select(SampleID, SubjectID, Cohort, Diet, Cat.12Mo, 
                     Percent.12Mo, Timepoint)) %>%
  filter(!is.na(Cat.12Mo) & !is.na(Cohort)) %>%
  mutate(Cohort = factor(Cohort, levels = c(2, 3), 
                         labels = c("Discovery", "Validation")))

pb_ratio_summary <- pb_ratio %>%
  group_by(SubjectID, Timepoint, Cohort, Cat.12Mo, Diet) %>%
  dplyr::summarise(meanPBratio = mean(Ratio), medianPBratio = median(Ratio), 
                   mean_logPBratio = mean(logRatio), median_logPBratio = median(logRatio)) 
head(pb_ratio_summary)
```

```{r}
pb_ratio_summary %>%
  filter(Timepoint == "Baseline") %>%
  group_by(Cohort) %>%
  summarise(frac_above = sum(meanPBratio > 0.01))
```

```{r pb-ratio-cohort-2-bySample, fig.height=6, fig.width=15}
pb.ratio_2 <- ggplot() +
    geom_boxplot(
        data = pb_ratio_summary %>% 
            filter(Timepoint == "Baseline", Cohort == "Discovery"), 
        aes(x = Cat.12Mo, y = mean_logPBratio, color = Diet),
        outlier.shape = NA) +
    geom_jitter(
        data = pb_ratio %>%
            filter(Timepoint == "Baseline", Cohort == "Discovery"), 
        aes(x = Cat.12Mo, y = logRatio), color = "grey55", size = 1.5, 
        width = 0.2
    ) + 
    geom_jitter(
        data = pb_ratio_summary %>%
            filter(Timepoint == "Baseline", Cohort == "Discovery"),
        aes(x = Cat.12Mo, y = mean_logPBratio, color = Diet),
        size = 4, height = 0, width = 0.1, shape = 17
    ) +
    # ggpubr::stat_compare_means(
    #     data = pb_ratio_summary %>% 
    #         filter(Timepoint == "Baseline",
    #                Cohort == "Discovery", Cat.12Mo != "MS"),
    #     aes(x = Cat.12Mo, y = mean_logPBratio),
    #     size = 6, label = "p.format", na.rm = TRUE) +
    stat_signif(
        data = pb_ratio_summary %>%
            filter(Timepoint == "Baseline", Cohort == "Discovery"),
        aes(x = Cat.12Mo, y = mean_logPBratio),
        stat = "signif",
        comparisons = list(c("VS", "US")),
        test = "wilcox.test",
        test.args = list(alternative = "two.sided"),
        margin_top = -0.15 , map_signif_level = FALSE,
        vjust = 9, hjust = -0.3, textsize = 6) +
    facet_wrap(~ Diet, scale = "free") +
    scale_color_manual(values = diet_cols) +
    labs(y = expression(paste(log[10], " (P/B)")), 
         x = "12-month weight loss success") + 
    theme(legend.position="none", 
          text = element_text(size = 30)) + 
    coord_flip() #+ ylim(-5.5,4.1)

plot_pb_2 <- ggplot_build(pb.ratio_2)
p.val <- plot_pb_2$data[[4]]$annotation
new_p.val <- as.factor(paste0("p=", p.val))
plot_pb_2$data[[4]]$annotation <- new_p.val

newPlot <- ggplot_gtable(plot_pb_2)

plot(newPlot)
```

```{r pb-ratio-cohort-3-bySample, fig.height=6, fig.width=15}
pb.ratio_3 <- ggplot() +
  geom_boxplot(data = pb_ratio_summary %>% filter(Timepoint == "Baseline", Cohort == "Validation"),
               aes(x = Cat.12Mo, y = mean_logPBratio, color = Diet), outlier.shape = NA) +
  geom_jitter(data = pb_ratio %>% filter(Timepoint == "Baseline", Cohort == "Discovery"), 
              aes(x = Cat.12Mo, y = logRatio), color = "grey66", size = 1.5, width = 0.2) + 
  geom_jitter(data = pb_ratio_summary %>% filter(Timepoint == "Baseline", Cohort == "Validation"),
              aes(x = Cat.12Mo, y = mean_logPBratio, color = Diet), size = 4, height = 0, width = 0.1, shape = 17) +
  stat_signif(data = pb_ratio_summary %>% filter(Timepoint == "Baseline", Cohort == "Validation"),
              aes(x = Cat.12Mo, y = mean_logPBratio),
              stat = "signif", 
              comparisons = list(c("VS", "US")),
              test = "wilcox.test", 
              test.args = list(alternative = "two.sided"),
               margin_top = -0.15, vjust = 9, hjust = -.3, textsize = 6) + #, 
  facet_wrap(~ Diet, scale = "free") +
  scale_color_manual(values = diet_cols) +
  labs(y = expression(paste(log[10], " (P/B)")), x = "12-month weight loss success") + 
  theme(legend.position="none", text = element_text(size = 30)) + coord_flip() #+ ylim(-5.5,4.1)

plot_pb_3 <- ggplot_build(pb.ratio_3)
p.val <- plot_pb_3$data[[4]]$annotation
new_p.val <- as.factor(paste0("p=", p.val))
plot_pb_3$data[[4]]$annotation <- new_p.val

newPlot <- ggplot_gtable(plot_pb_3)

plot(newPlot)
```


```{r}
wtest_pb <- pb_ratio_summary %>%
  ungroup() %>%
  filter(Cat.12Mo %in% c("US", "VS")) %>%
  mutate(Cat.12Mo = factor(Cat.12Mo, levels = c("US", "VS"), ordered = TRUE)) %>%
  group_by(Cohort, Timepoint, Diet) %>%
  data.frame() %>%
  do(broom::tidy(wilcox.test(formula = mean_logPBratio ~ Cat.12Mo, data = (.))))
wtest_pb
```




```{r}
table(wtest_pb$Cat.12Mo)
```


# Alpha-Diversity

## Discovery Cohort

### Low-Carb diet
```{r}
blueBL <- subset_samples(ps.bl, Color == "Blue" & Cohort == 2)
blueBL <- prune_taxa(taxa_sums(blueBL) > 0, blueBL)
blueBL

blueBL_cnts <- as(blueBL@otu_table, "matrix")
blueBL_cnts <- data.frame(t(blueBL_cnts))
blueBL_smp <- data.frame(blueBL@sam_data)
```


```{r, eval = FALSE}
# rarelevel1 <- unique(floor(10^(seq(2, 4, length.out = 20))/100)*100)
# blue_curves <- lapply(1:nsamples(blueBL), function(i) {
#   cat("Sample no.", i, "\n")
#   icounts <- blueBL_cnts[i, ]
#   stepRare <- c(rarelevel1, seq(11000, sum(icounts), by = 1000))
#   icurve <- phylocurve(icounts, blueBL@phy_tree, stepm = stepRare,
#                        subsampling = "individual", replace = TRUE,
#                        verbose = FALSE)
#   df <- data.frame(m = icurve[, 1], pdrare = icurve[, 2],
#                    pdvar = icurve[, 3], blueBL_smp[rep(i, nrow(icurve)), ])
#   return(df)
# })
# blue_curves_df <- do.call("rbind", blue_curves)
# saveRDS(blue_curves_df, file = "FINAL ANALYSIS/data/alphaDiversity/diet_phylocurves_blue_cohort2.rda")

#load("FINAL ANALYSIS/data/alphaDiversity/diet_phylocurves.rda")
```

```{r}
blue_curves_df <- readRDS("../data/alphaDiversity/diet_phylocurves_blue_cohort2.rda")
```




```{r}
plot_alphadiver_curves <- function(df, cutoff = 11000, color = "Percent.12Mo"){
  ggplot(df, aes_string(group = "SampleID", color = color, x = "m", y = "pdrare")) + 
    geom_line(lwd = 0.8) + 
    geom_line(aes(y = pdrare + sqrt(pdvar)), lty = 3, lwd = 0.3) + 
    geom_line(aes(y = pdrare - sqrt(pdvar)), lty = 3, lwd = 0.3) + 
    scale_x_log10(breaks = c(100, 1000, 10000, 100000)) +
    ylab("Expected Phylogenetic Diversity") + xlab("Rarefaction size") +
    geom_vline(xintercept = cutoff) +
    scale_color_viridis(discrete = !is.numeric(df[[color]])) +
    theme(text = element_text(size = 20))
}

plot_alphadiver_boxplot <- function(df, cutoff = 11000){
  df <- df %>% filter(floor(m) == cutoff)
  ggplot(df, aes(x = Cat.12Mo, y = pdrare)) +
    geom_boxplot() + scale_color_viridis() +
    geom_jitter(aes(color = Percent.12Mo), width = 0.3, size = 1) +
    stat_signif(
      aes(y = pdrare, x = Cat.12Mo),
      stat = "signif", comparisons = list(c("VS", "US")), 
      test = "wilcox.test", test.args = list(alternative = "two.sided"), 
      manual = FALSE, y_position = 30, textsize = 5) + 
    ylab("Expected Phylogenetic Diversity") + xlab("12-month weight loss success category") +
    theme(text = element_text(size = 20))
}



plot_alphadiver <- function(
  df, cutoff = 11000, vp = grid::viewport(
    width = 0.3, height = 0.32, x = 0.1, y = 0.76, just = c("left", "top"))) {
  p0 <- plot_alphadiver_curves(df, cutoff = cutoff, color = "Percent.12Mo") + ylim(0,36) + labs(color = "Percent weight loss at 12-months") + theme(legend.position = "top", legend.title=element_text(size=14)) + guides(color = guide_colourbar(barwidth = 15, title.position = "top"))
  p1 <- plot_alphadiver_boxplot(df, cutoff = cutoff) + 
  guides(color = "none") + xlab("") + ylab("") +
  theme(panel.background = element_rect(fill = "transparent",colour = NA),
        plot.background = element_rect(fill = "transparent",colour = NA),
        text = element_text(size = 15)) + ylim(5, 33) 
  print(p0)
  theme_set(theme_bw(base_size = 8))
  print(p1, vp = vp)
  theme_set(theme_bw())
}

```



```{r phylocurves-blue-bl-per12-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df) 
```

```{r}
blue_curves_df <- blue_curves_df %>%
  left_join(blueBL_smp %>%
              select(SampleID, adherence.changeFromBL, adherence.pct), 
            by = c("SampleID"))
```

```{r phylocurves-blue-bl-reduction-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df, color = "adherence.changeFromBL") +
  scale_color_viridis(option = "inferno") 
```

```{r phylocurves-blue-bl-adherence-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df, color = "adherence.pct") +
  scale_color_viridis() 
```



```{r phylocurves-blue-bl-gender-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df, color = "Gender") +
  scale_color_brewer(palette = "Dark2") 
```


```{r alphadiv-blue-bl-cohort2, fig.width=7, fig.height=6}
plot_alphadiver(blue_curves_df)
```


```{r}
(blue_PD_summary_cohort2 <- blue_curves_df %>%
   filter(floor(m) == 11000) %>%
   group_by(Cat.12Mo) %>%
   summarise(mean(pdrare), sd(pdrare)))
  
with(blue_curves_df %>%
   filter(floor(m) == 11000), wilcox.test(
  pdrare[Cat.12Mo == "US"], pdrare[Cat.12Mo == "VS"]))

```






### Low-Fat diet


```{r}
purpBL <- subset_samples(ps.bl, Color == "Purple" & Cohort == 2)
purpBL <- prune_taxa(taxa_sums(purpBL) > 0, purpBL)
purpBL

purpBL_cnts <- as(purpBL@otu_table, "matrix")
purpBL_cnts <- data.frame(t(purpBL_cnts))
purpBL_smp <- data.frame(purpBL@sam_data)
```


```{r}
# purp_curves <- lapply(1:nsamples(purpBL), function(i) {
#   cat("Sample no.", i, "\n")
#   icounts <- purpBL_cnts[i, ]
#   stepRare <- c(rarelevel1, seq(11000, sum(icounts), by = 1000))
#   icurve <- phylocurve(icounts, purpBL@phy_tree, stepm = stepRare,
#                        subsampling = "individual", replace = TRUE,
#                        verbose = FALSE)
#   df <- data.frame(m = icurve[, 1], pdrare = icurve[, 2],
#                    pdvar = icurve[, 3], purpBL_smp[rep(i, nrow(icurve)), ])
#   return(df)
# })
# purp_curves_df <- do.call("rbind", purp_curves)
# save(list  = c("blue_curves_df", "purp_curves_df"), file = "diet_phylocurves_cohort2.rda")
# saveRDS(purp_curves_df, file = "FINAL ANALYSIS/data/alphaDiversity/diet_phylocurves_purple_cohort2.rda")


purp_curves_df <- readRDS("../data/alphaDiversity/diet_phylocurves_purple_cohort2.rda")
```

```{r}
alphadiver_cohort2 <- bind_rows(blue_curves_df, purp_curves_df) %>%
  filter(floor(m) == 11000)
```



```{r phylocurves-purp-bl-per12-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df) 
```




```{r}
purp_curves_df <- purp_curves_df %>%
  left_join(purpBL_smp %>%
              select(SampleID, adherence.changeFromBL, adherence.pct), 
            by = c("SampleID"))
```

```{r phylocurves-purp-bl-reduction-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "adherence.changeFromBL") +
  scale_color_viridis(option = "inferno") 
```

```{r phylocurves-purp-bl-adherence-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "adherence.pct") +
  scale_color_viridis() 
```


```{r phylocurves-purp-bl-age-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "Age") +
  scale_color_viridis(option = "inferno", end = 0.85) 
```

```{r phylocurves-purp-bl-gender-cohort2, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "Gender") +
  scale_color_brewer(palette = "Set2") 
```


```{r alphadiv-purp-bl-cohort2, fig.width=7, fig.height=6}
plot_alphadiver(purp_curves_df)
```



```{r}
(purple_PD_summary_cohort2 <- purp_curves_df %>%
   filter(floor(m) == 11000) %>%
   group_by(Cat.12Mo) %>%
   summarise(mean(pdrare), sd(pdrare)))
  
with(purp_curves_df %>%
   filter(floor(m) == 11000), wilcox.test(
  pdrare[Cat.12Mo == "US"], pdrare[Cat.12Mo == "VS"]))

```

## Validation Cohort

### Low-Carb diet
```{r}
blueBL <- subset_samples(ps0, Timepoint == "Baseline" & Color == "Blue" & Cohort == 3)
blueBL <- prune_taxa(taxa_sums(blueBL) > 0, blueBL)
blueBL

blueBL_cnts <- as(blueBL@otu_table, "matrix")
blueBL_cnts <- data.frame(t(blueBL_cnts))
blueBL_smp <- data.frame(blueBL@sam_data)

```

```{r}
#rarelevel1 <- unique(floor(10^(seq(2, 4, length.out = 20))/100)*100)
# blue_curves <- lapply(1:nsamples(blueBL), function(i) {
#   cat("Sample no.", i, "\n")
#   icounts <- blueBL_cnts[i, ]
#   stepRare <- c(rarelevel1, seq(11000, sum(icounts), by = 1000))
#   icurve <- phylocurve(icounts, blueBL@phy_tree, stepm = stepRare,
#                        subsampling = "individual", replace = TRUE,
#                        verbose = FALSE)
#   df <- data.frame(m = icurve[, 1], pdrare = icurve[, 2],
#                    pdvar = icurve[, 3], blueBL_smp[rep(i, nrow(icurve)), ])
#   return(df)
# })
# blue_curves_df <- do.call("rbind", blue_curves)
# saveRDS(blue_curves_df, file = "FINAL ANALYSIS/data/alphaDiversity/diet_phylocurves_blue_cohort3.rda")

#load("FINAL ANALYSIS/data/alphaDiversity/diet_phylocurves.rda")
blue_curves_df <- readRDS("../data/alphaDiversity/diet_phylocurves_blue_cohort3.rda")
```

```{r phylocurves-blue-bl-per12-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df) 
```



```{r}
blue_curves_df <- blue_curves_df %>%
  left_join(blueBL_smp %>%
              select(SampleID, adherence.changeFromBL, adherence.pct), 
            by = c("SampleID"))
```

```{r phylocurves-blue-bl-reduction-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df, color = "adherence.changeFromBL") +
  scale_color_viridis(option = "inferno") 
```

```{r phylocurves-blue-bl-adherence-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df, color = "adherence.pct") +
  scale_color_viridis() 
```



```{r phylocurves-blue-bl-age-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df, color = "Age") +
  scale_color_viridis(option = "inferno", end = 0.85) 
```

```{r phylocurves-blue-bl-gender-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(blue_curves_df, color = "Gender") +
  scale_color_brewer(palette = "Set2") 
```

```{r alphadiv-blue-bl-cohort3, fig.width=7, fig.height=6}
plot_alphadiver(blue_curves_df)
```


```{r}
(blue_PD_summary_cohort3 <- blue_curves_df %>%
   filter(floor(m) == 11000) %>%
   group_by(Cat.12Mo) %>%
   summarise(mean(pdrare), sd(pdrare)))
  
with(blue_curves_df %>%
   filter(floor(m) == 11000), wilcox.test(
  pdrare[Cat.12Mo == "US"], pdrare[Cat.12Mo == "VS"]))

```



### Low-Fat diet


```{r}
purpBL <- subset_samples(ps0, Timepoint == "Baseline" & Color == "Purple" & Cohort == 3)
purpBL <- prune_taxa(taxa_sums(purpBL) > 0, purpBL)
purpBL

purpBL_cnts <- as(purpBL@otu_table, "matrix")
purpBL_cnts <- data.frame(t(purpBL_cnts))
purpBL_smp <- data.frame(purpBL@sam_data)
```


```{r}
# purp_curves <- lapply(1:nsamples(purpBL), function(i) {
#   cat("Sample no.", i, "\n")
#   icounts <- purpBL_cnts[i, ]
#   stepRare <- c(rarelevel1, seq(11000, sum(icounts), by = 1000))
#   icurve <- phylocurve(icounts, purpBL@phy_tree, stepm = stepRare,
#                        subsampling = "individual", replace = TRUE,
#                        verbose = FALSE)
#   df <- data.frame(m = icurve[, 1], pdrare = icurve[, 2],
#                    pdvar = icurve[, 3], purpBL_smp[rep(i, nrow(icurve)), ])
#   return(df)
# })
# purp_curves_df <- do.call("rbind", purp_curves)
# save(list  = c("blue_curves_df", "purp_curves_df"), file = "diet_phylocurves_cohort3.rda")
# saveRDS(purp_curves_df, file = "FINAL ANALYSIS/data/alphaDiversity/diet_phylocurves_purple_cohort3.rda")


purp_curves_df <- readRDS("../data/alphaDiversity/diet_phylocurves_purple_cohort3.rda")
```



```{r phylocurves-purp-bl-per12-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df)
```


```{r}
purp_curves_df <- purp_curves_df %>%
  left_join(purpBL_smp %>%
              select(SampleID, adherence.changeFromBL, adherence.pct), 
            by = c("SampleID"))
```

```{r phylocurves-purp-bl-reduction-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "adherence.changeFromBL") +
  scale_color_viridis(option = "inferno") 
```

```{r phylocurves-purp-bl-adherence-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "adherence.pct") +
  scale_color_viridis() 
```


```{r phylocurves-purp-bl-age-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "Age") +
  scale_color_viridis(option = "inferno", end = 0.85) 
```

```{r phylocurves-purp-bl-gender-cohort3, fig.width=10, fig.height=6}
plot_alphadiver_curves(purp_curves_df, color = "Gender") +
  scale_color_brewer(palette = "Set2") 
```


```{r alphadiv-purp-bl-cohort3, fig.width=7, fig.height=6}
plot_alphadiver(purp_curves_df)
```



```{r}
(purple_PD_summary_cohort3 <- purp_curves_df %>%
   filter(floor(m) == 11000) %>%
   group_by(Cat.12Mo) %>%
   summarise(mean(pdrare), sd(pdrare)))
  
with(purp_curves_df %>%
   filter(floor(m) == 11000), wilcox.test(
  pdrare[Cat.12Mo == "US"], pdrare[Cat.12Mo == "VS"]))

```







```{r}
sessionInfo()
```