Bio720_R_InClass_Control_flow_fundamentals.Rmd

---
title: "Control flow in R, the fundamentals"
author: "Ian Dworkin"
date: "`r format(Sys.time(),'%d %b %Y')`"
output: 
  html_document: 
    toc: yes
    number_sections: yes
    keep_md: yes
editor_options: 
  chunk_output_type: console
---


# Worked examples using control flow

```{r}
knitr::opts_chunk$set(eval = FALSE)
```

**NOTE:** I have set the defaults (line above) for evaluation of blocks of code to be off (`eval = FALSE`). When you are running the code you might need to change these. I did this for rendering this activity.


## if else

Here is an example of using `if` and `else` to do different computation (or in this case, different output) depending on the input

```{r}
p_test <- function(p){
    if (p <= 0.05) print ("yeah!!!!") 
        else if (p >= 0.9) print ("high!!!!")
        else print ("somewhere in the middle")
}
```


Try some numbers
```{r}
p_test(0.5)
```


Try a random number from a uniform distribution on $[0,1]$ using `runif`

```{r}
(test_val_1 <- runif(n = 1))
p_test(test_val_1)
```


But we often don't have just a single number, but a vector of numbers we need to evaluate and make decisions with.  Let's try it with 100 random numbers

```{r}
(test_val_2 <- runif(n = 100))
```


```{r}
p_test(test_val_2)
```


What does it not like?

The `if` and `else` function are not naturally vectorized. 

What are our options?

(`for` loop, `sapply`, `ifelse`)

```{r}
p_test_2 <- function(p){
    ifelse(test = p <= 0.05, 
           yes = print("yippee"), 
           no = print("bummer")) }
```

```{r}
p_test_2(test_val_2)
```

How many times were we happy or sad
```{r}
table(p_test_2(test_val_2))
```

For genomics we do sometimes need to assess things just like this, how many of our genomic features "pass" some threshold
```{r}
feature_tests <- runif(n = 17300, min = 0, max = 1)
```


via for loops
```{r}
p_vals_out <- rep(NA, length(feature_tests))

for (i in 1:length(p_vals_out)) {
  p_vals_out[i] <- p_test(feature_tests[i])
}
```


via sapply
```{r}
p_vals_sapply_out <- sapply(feature_tests, p_test)
```

As you can see, we really should rewrite the functions so they are not outputting things to the screen


```{r}
p_test_V2 <- function(p){
    if (p <= 0.05) x = "yeah!!!!"
        else if (p >= 0.9) x = "high!!!!"
        else x = "somewhere in the middle"
    return(x)    
}

p_test_V2(0.5)
```

clean output!
```{r}
p_vals_sapply_out <- sapply(feature_tests, p_test_V2)
```


```{r}
p_test_3 <- function(p){
    ifelse(test = p <= 0.05, 
           yes = "yippee", 
           no =  "bummer") }

p_test_3(feature_tests)
```