4.4-Processing_lists.qmd

---
title: "Working with Lists of Plots"
author:
  - Elizabeth King
  - Kevin Middleton
format:
  revealjs:
    theme: [default, custom.scss]
    standalone: true
    self-contained: true
    logo: QMLS_Logo.png
    slide-number: true
    show-slide-number: all
    link-external-newwindow: true
---

## What else can we do?

```{r}
#| label: setup
#| echo: false
#| warning: false
#| message: false

library(tidyverse)
library(cowplot)
ggplot2::theme_set(theme_cowplot())
library(patchwork)
```

- Anything else you want in the function
  - Summary stats
  - Regression
  - etc.
  

## Mammal life history data

```{r}
M <- read_delim("Data/mammals.txt", show_col_types = FALSE) |> 
  mutate(across(.cols = where(is.double), 
                .fns = ~ ifelse(.x == -999, NA, .x))) |> 
  rename(Order = MSW05_Order,
         Species = MSW05_Binomial,
         Body_mass = `5-1_AdultBodyMass_g`,
         Forearm = `8-1_AdultForearmLen_mm`,
         Body_length = `13-1_AdultHeadBodyLen_mm`,
         Eye_opening = `2-1_AgeatEyeOpening_d`,
         Repro_age = `3-1_AgeatFirstBirth_d`,
         BMR_1 = `18-1_BasalMetRate_mLO2hr`,
         BMR_2 = `5-2_BasalMetRateMass_g`,
         Neonate_mass = `5-3_NeonateBodyMass_g`) |> 
  select(Order, Species, Body_mass, Forearm, Body_length, Eye_opening,
         Repro_age, BMR_1, BMR_2, Neonate_mass)
head(M)
```


## Update processing function

```{r}
#| echo: true
trait_analysis <- function(trait, M) {
  M_no_NA <- M |> 
    drop_na(Order, Body_mass, all_of(trait)) |> 
    select(Order, Body_mass, all_of(trait)) |> 
    mutate(across(.cols = where(is.double), .fns = ~ log10(.x + 0.001)))
  
  summary_table <- M_no_NA |> 
    group_by(Order) |> 
    summarize(across(.cols = where(is.double),
                     .fns = mean))
  
  p <- ggplot(M_no_NA, aes(x = Body_mass, y = .data[[trait]], color = Order)) +
    geom_point() +
    labs(y = str_replace(trait, "_", " ")) +
    theme(legend.position = "none")
  
  fm <- lm(M_no_NA[[trait]] ~ M_no_NA[["Body_mass"]])
  
  return(list(trait = trait, p = p, summary_table = summary_table, fm = fm))
}
```


## Process traits

- Pass the new function as `.f`

```{r}
#| echo: true
LL <- map(.x = names(M)[4:10],
          .f = trait_analysis,
          M = M)
```


## A list of lists

Each element in the list has

1. Trait (string)
2. Plot
3. Table of means
4. An `lm()` fitted model

```{r}
#| echo: true
length(LL)
names(LL[[1]])
```


## Access parts of the list

```{r}
#| echo: true
LL[[1]]$summary_table
```


## What can we do now?

- Extract all the plots or tables or regressions
- Generate a report with each trait in turn

## Extract all the plots

```{r}
plots <- map(.x = 1:length(LL),
             .f = function(ii, LL) {
               return(LL[[ii]]$p)
             },
             LL = LL)
wrap_plots(plots)
```

`pluck(LL, ii, "p")` would also work here.


## Generate a report with each trait

In a Quarto file:

```{{r}}
#| output: "asis"
for (ii in 1:length(LL)) {

  # Make a header
  cat("\n## ", LL[ii], "\n")
  
  # Print formatted table
  knitr::kable(LL[[ii]]$summary_table)
  
  # Print plot
  print(LL[[ii]]$p)
  
  # Print linear model summary
  print(summary(LL[[ii]])$fm)
}
```