---
title: "scicalc"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{scicalc}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

```{r, include=FALSE}
library(dplyr)
data_source <- file.path(here::here(), "inst", "extdata", "data", "source")

for (file in c("dm.sas7bdat", "vs.parquet", "lb.csv", "pc.parquet")) {
  unlist(file.path(data_source, file))
}

df <- data.frame(
  "ID" = factor(c(1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3)),
  "SEX" = c("F", "F", "F", "F", "M", "M", "M", "M", "F", "F", "F", "F"),
  "RACE" = c("WHITE", "WHITE", "WHITE", "WHITE", "BLACK", "BLACK", "BLACK", "BLACK", "ASIAN", "ASIAN", "ASIAN", "ASIAN"),
  "AGE" = c(24, 24, 24, 24, 22, 22, 22, 22, 35, 35, 35, 35),
  "CREAT" = c(1, 1, 1, 1, 4, 4, 4, 4, 3, 3, 3, 3),
  "CREATU" = c("mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL"),
  "CYSTC" = c(0.4, 0.4, 0.4, 0.4, 0.9, 0.9, 0.9, 0.9, 0.7, 0.7, 0.7, 0.7),
  "CYSTCU" = c("mg/L","mg/L","mg/L","mg/L","mg/L","mg/L","mg/L","mg/L","mg/L","mg/L","mg/L","mg/L"),
  "AST" = c(15, 15, 15, 15, 29, 29, 29, 29, 38, 38, 38, 38),
  "ULNAST" = c(33, 33, 33, 33, 33, 33, 33, 33, 33, 33, 33, 33),
  "ASTU" = c("U/L","U/L","U/L","U/L","U/L","U/L","U/L","U/L", "U/L", "U/L", "U/L", "U/L"),
  "BILI" = c(0.8, 0.8, 0.8, 0.8, 1.9, 1.9, 1.9, 1.9, 1.1, 1.1, 1.1, 1.1),
  "ULNBILI" = c(1.2, 1.2, 1.2, 1.2, 1.2, 1.2, 1.2, 1.2, 1.2, 1.2, 1.2, 1.2),
  "BILIU" = c("mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL","mg/dL"),
  "HEIGHT" = c(174 ,174, 174, 174, 201, 201, 201, 201, 190, 190, 190, 190),
  "WEIGHT" = c(70, 70, 70, 70, 80, 80, 80, 80, 75, 75, 75, 75),
  "DV" = c(10, 150, 70, 9, 7, 140, 60, 7, 11, 156, 70, 7),
  "NTLD" = c(0, 0.25, 1, 8, 0, 0.25, 1, 8, 0, 0.25, 1, 8)
)

library(ggplot2)
ggplot2::ggplot(df, aes(x = NTLD, y = DV, color = ID)) +
  geom_point() +
  geom_line()

df %>% 
  dplyr::select(c("ID", "SEX", "RACE", "AGE")) %>% 
  haven::write_sas(path = file.path(data_source, "dm.sas7bdat"))

df %>% 
  dplyr::select(c("ID", "HEIGHT", "WEIGHT")) %>% 
  arrow::write_parquet(sink = file.path(data_source, "vs.parquet"))

df %>% 
  dplyr::select(c("ID", "CREAT", "CREATU", "CYSTC", "CYSTCU", "AST", "ULNAST", "BILI", "ULNBILI", "ASTU", "BILIU")) %>% 
  readr::write_csv(file = file.path(data_source, "lb.csv"))

df %>% 
  dplyr::select(c("ID", "DV", "NTLD")) %>% 
  arrow::write_parquet(sink = file.path(data_source, "pc.parquet"))

```

To install the scicalc package you can run the following two commands:
```r
options(repos = c('scicalc' = "https://a2-ai.github.io/gh-pkg-mirror/scicalc",
                  getOption("repos")))
install.packages("scicalc")
```
# Scicalc
Scicalc is a package of functions commonly used at A2-Ai. We hope this package will ease data assembly and allow for a more reproducible code across analysts and projects. It is in no way complete or all encompassing, but we hope this is a good start. Please let us know if there are additional functions we should include!
```{r setup}
library(scicalc)
```
## Reading in data

I'll first try and read in data from `~/inst/extdata/data/source` - in R packages extra files go in the `inst` directory, but we can pretend this is just like a Project starter `data/source` directory.
```{r}
data_source <- file.path(here::here(), "inst", "extdata", "data", "source")
if (!dir.exists(data_source)) {
  fs::dir_create(data_source)
}

list.files(data_source)
```
This is a contrived example, but there are several different file types in source data representing different SDTM domains. As well as a `test.txt` file. We can use the `read_file_with_hash` function to read in all the files in this source directory. No more creating `read_file_type_with_hash` functions in each script!


```{r}
raw_data <- purrr::map(
  list.files(data_source, full.names = TRUE),
  scicalc::read_file_with_hash) %>% 
  purrr::set_names(tools::file_path_sans_ext(list.files(data_source)))
```
These hashes come from `digest::digest`'s default algorithm `md5`, but we can alter this by supplying `algo = 'blake3'` to the read_file_with_hash function.
```{r}
raw_data <- purrr::map(
  list.files(data_source, full.names = TRUE),
  scicalc::read_file_with_hash, algo = "blake3") %>% 
  purrr::set_names(tools::file_path_sans_ext(list.files(data_source)))
```


```{r}
names(raw_data)
```

We can see that each different file type was read with read_file_with_hash and saved to `raw_data`. We also got a warning for `test.txt` file was not currently supported. If there are any other file types you commonly read and want to print a hash besides `csv`, `parquet`, and `sas7bdat` please reach out so we can add that functionality! Behind the scenes of `read_file_with_hash` the extension of the file is determined and a corresponding function such as `read_parquet_with_hash` is called. You can call either, but I would recommend using the general `read_file_with_hash`.

## Simple data analysis

Let's do some bmi and bsa caclulations with the data in the vs domain
```{r}
df <- raw_data$pc

df <- df %>% 
  dplyr::mutate(.data = raw_data$vs,
    BBMI = scicalc::bbmi(WEIGHT, HEIGHT),
    DBSA = scicalc::bsa(WEIGHT, HEIGHT),
    MBSA = scicalc::bsa(WEIGHT, HEIGHT, method = "mosteller"))
df
```

Now we can verify the lab units before moving on - remember this is a contrived dataset, but the workflow should be somewhat similar.
```{r}
dflb <- raw_data$lb %>% 
  dplyr::select(c("ID", "CREAT", "CYSTC", "AST", "BILI", )) %>% 
  dplyr::distinct() %>% 
  tidyr::pivot_longer(cols = c("CREAT", "CYSTC", "AST", "BILI"), 
                      names_to = "LBTESTCD", values_to = "LBORRES")

dfu <- raw_data$lb %>%
  dplyr::select(c("ID", "CREATU", "CYSTCU", "ASTU", "BILIU")) %>% 
  dplyr::distinct() %>% 
  tidyr::pivot_longer(cols = c("CREATU", "CYSTCU", "ASTU", "BILIU"), 
                      names_to = "LBTESTCDU", values_to = "LBORRESU") 

lb <- dflb %>% 
  dplyr::mutate(LBORRESU = dfu$LBORRESU)

lb
```
```{r}
if (scicalc::check_for_unique_units(lb$LBTESTCD, lb$LBORRESU)) {
  print("Units are 1:1!")
} else {
  stop("Error with units!")
}
```
```{r}
scicalc::get_unique_units_df(lb$LBTESTCD, lb$LBORRESU)
```
Now we can use the lab data to calculate hepatic function.
```{r}
df <- df %>% 
  dplyr::mutate(.data = raw_data$lb, 
    BHFC = scicalc::bhfc(AST, ULNAST, BILI, ULNBILI),
  )
df
```
Now we'll need the demographics needed for Creatinine clearance, estimated glomerular filtration rate, and renal function.
```{r}
df <- df %>% 
  dplyr::mutate(.data = raw_data$dm, 
    CRCL = scicalc::crcl(scicalc::is_female(SEX), AGE, CREAT, WEIGHT),
    BRFC = scicalc::brfc(CRCL),
    ckdepi_2009_egfr = scicalc::egfr(
      sexf = scicalc::is_female(SEX), 
      raceb = scicalc::is_black(RACE), 
      age = AGE, 
      creat = CREAT,
      method = "CKDEPI 2009"),
    ckdepi_2021_egfr = scicalc::egfr(
      sexf = scicalc::is_female(SEX),
      age = AGE,
      creat = CREAT,
      method = "CKDEPI 2021"),
    ckdepi_2021_egfr_cystatin = scicalc::egfr(
      sexf = scicalc::is_female(SEX), 
      age = AGE, 
      creat = CREAT, 
      cystc = CYSTC, 
      method = "CKDEPI 2021 CYSTATIN"),
    mdrd_egfr = scicalc::egfr(
      sexf = scicalc::is_female(SEX), 
      raceb = scicalc::is_black(RACE), 
      age = AGE, 
      creat = CREAT,
      method = "MDRD"),
    schwartz_egfr = scicalc::egfr(
      height = HEIGHT,
      creat = CREAT,
      method = "Schwartz"
    )
  )

df
```
## Writting data
We can use the `write_file_with_hash` to write `parquet` files (We highly recommend you write all data to parquet unless there is a specific need for other file type). We just need to supply a path to the file if the directory the file should be written to does not exist, this function will create it and all directories needed on the path. This function won't let us overwrite the file unless we supply `overwrite = TRUE`. Like `read_file_with_hash` we can choose which algorithm to use for the digest.
```{r}
data_derived <- file.path(here::here(), "inst", "extdata", "data", "derived")
if (!file.exists(file.path(data_derived, "pk_data_v01.parquet"))) {
  scicalc::write_file_with_hash(
    df, file.path(data_derived, "pk_data_v01.parquet"), algo = "blake3")  
} else {
  print("Overwriting data!")
  scicalc::write_file_with_hash(
    df, file.path(data_derived, "pk_data_v01.parquet"), overwrite = TRUE, algo = "blake3")
}
```
## Reading in derived data
Now if we want to read in the derived data and verify it is the data we think it is we can use `read_hashed_file`. This function will only let you ingest the data if the hash you supply matches the hash of the contents of the file. Here I've supplied the `blake3` hash which does not match the `md5` hash digest uses by default. We have two options, tell `read_hashed_file` to use `algo = 'blake3'` or supply the `md5` hash.
```{r}
hash <- digest::digest(file.path(data_derived, "pk_data_v01.parquet")) # This is a hash of the string to the file type so it will not match the hash for contents of the file.
print(hash)
#This hash is incorrect so the function errors
if (file.exists(file.path(data_derived, "pk_data_v01.parquet"))) {
  testthat::expect_error(scicalc::read_hashed_file(file.path(data_derived, "pk_data_v01.parquet"), hash))
}
```
1. updated algo argument works:
```{r}
correct_hash <- digest::digest(file = file.path(data_derived, "pk_data_v01.parquet"), algo = "blake3")
print(correct_hash)
```
```{r}
data <- scicalc::read_hashed_file(
  file.path(data_derived, "pk_data_v01.parquet"),
  algo = "blake3",
  correct_hash)
data
```

2. giving the `md5` hash works
```{r}
md5_hash <- digest::digest(file = file.path(data_derived, "pk_data_v01.parquet"))
print(md5_hash)
```
```{r}
data <- scicalc::read_hashed_file(
  file.path(data_derived, "pk_data_v01.parquet"), 
  md5_hash)
data
```
```{r, include=FALSE}

```