fill_tablelookup.qmd

# Lookup 

```{r}
#| results: "asis"
#| echo: false
source("_common.R")
```

## Introduction

Functions from the Tidyverse packages (`dplyr` and `tidyr`) will be used in this tutorial. As example data the lipid names of a lipidomics dataset is being.

```{r}
#| label: load-lib
#| echo: true
#| message: false
library(here)
library(tidyverse)
library(glue)

# Make some test data

my_database <- tibble(
  ID = 1:7, 
  lipid = c("CE 16:0", "Cer d18:1/16:0", "Hex3Cer d18:1/22:0", "LPC 17:0", "LPC O-18:0", "Cer 32:1;2", "LPE 20:0"),
  MZ = c(642.6071, 538.5020,1108.8231, 510.355416, 510.391802, 510.488071, 510.355416)
)

my_results <- tribble(
  ~FeatureNo, ~mz,
  1, 1108.8241,
  2, 642.607,
  3, 642.657,
  4, 642.87,
  5, 538.5021,
  6, 510.391802,
  7, 510.390902,
  8, 510.32822,
  9, 123.32,
  10, 0, 
  11, NA,
  12, -1313.12
)

my_database
my_results
```

```{r}
#| label: select-4
#| echo: true
#| message: false
#| eval: false

# Define function to search database (takes 1 value, returns 1 value, or NA + warnings)

get_feature <- function(database, mz, ppm) {
  if(mz <= 0 |is.na(mz)){
    cat("Invalid mz\n")
    return(NA)
  }
  
  res <- database |> filter(MZ > mz - mz * ppm/1E6, MZ < mz + mz * ppm/1E6) |> pull(lipid)
  
  if (length(res) == 0){
    print(glue("Feature m/z={mz} has no match"))
    return(NA)
  } else if (length(res) > 1) {
    matches <- paste0(res, collapse = "; ")
    print(glue("Feature m/z={mz} has {length(res)} matches: {matches}"))
    return(NA)
  }
  res
}

# some first tests of the function

get_feature(my_database, 1108.83, 1)
get_feature(my_database, 1108.83, 11)
get_feature(my_database, 1108.83, 1000000)
get_feature(my_database, NA)
get_feature(my_database, NA, 11)

```

```{r}

# use function on a dataset

my_annotated_results <- my_results |> 
  rowwise() |>   # do not forget this 
  mutate(lipid = get_feature(database = my_database, mz = mz, ppm = 100))

```