DATADATADATA: Data Hoarding

Increasing the Value of Data Requires Activation!

Data Driven Decision Making

Because we’ve always done it this way!

Levering data driven decision making allows the company to gain a competitive edge and this is where you Bio Data Science skills are indispensable!

You value as a Bio Data Scientist / Bioinformatician

R for Bio Data Science - What is it?

• In essence: The art of converting numbers to value
  • Ingest data
  • Transform, wrangle, visualise, model
  • Output insights

R for Bio Data Science - Intrinsically interdisciplinary

Why “Bio” in R for Bio Data Science?

What is the motivation for this Course?

What is the motivation for this Course?

What will you learn?

• The craft of going from raw extracted data to insights
• Advanced data visualisation
• Collaborative project oriented coding
• All with an emphasis on reproducibility and communication

Introducing R: A Journey into Bio Data Science

“To understand computations in R, two slogans are helpful: Everything that exists is an object. Everything that happens is a function call.” – John Chambers (creator of the S language, of which R is an implementation).

The Roots and Rise of R

A Few Examples of Functional Programming

my_vector <- c(49, 31, 24, 35, 71, 7, 36, 23, 67, 37)

Vector <- R6::R6Class("Vector",
  public = list(
    data = NULL,
    initialize = function(data) {
      if (!is.numeric(data)) {
        stop("Data should be numeric.")
      }
      self$data <- data
    },
    mean = function() {
      return(sum(self$data) / length(self$data))
    }
  )
)
numbers <- Vector$new(my_vector)
print(numbers$mean())

[1] 38

A Few Examples of Functional Programming

my_vector <- c(49, 31, 24, 35, 71, 7, 36, 23, 67, 37)

my_sum <- 0
for( i in 1:length(my_vector) ){
  my_sum <- my_sum + my_vector[i]
}
my_mean <- my_sum / length(my_vector)
print(my_mean)

[1] 38

A Few Examples of Functional Programming

my_vector <- c(49, 31, 24, 35, 71, 7, 36, 23, 67, 37)

my_mean <- mean(my_vector)
print(my_mean)

[1] 38

You simply call functions on objects

Standard Deviation

sd(my_vector)

[1] 19.70899

Median

median(my_vector)

[1] 35.5

Permute

sample(my_vector)

 [1] 31 23 67 35 49  7 71 37 36 24

Bootstrap

sample(my_vector, replace = TRUE)

 [1] 37 23 49  7 31 35 36 23 31 35

…and tons more!

“R is not a real programming language”: Debunking Myths I

1. R is Turing-complete:
   • R can theoretically solve any computational problem. Foundational concept shared with e.g. Python, C++, Java, etc.
2. R is fully capable in Production:
   • E.g. shiny apps used in industry and R comes with an ecosystem supporting reproducibility in production settings.
3. Comprehensive Ecosystem:
   • CRAN contains ~20,000 packages. Also Bioconductor is a gold standard for bioinformatics software.
4. Interoperability:
   • Seamless integration with other languages (C, C++, Fortran, and Python) using packages like Rcpp and reticulate.

“R is not a real programming language”: Debunking Myths II

5. Advanced Programming Features:
   • Supports object-oriented, functional, and imperative programming paradigms. Flexible metaprogramming with capabilities like non-standard evaluation
6. R’s Active & Growing Community:
   • Annual global R conferences and numerous local user groups and also: tidyverse
7. Performance:
   • R is interpreted and can be slower, packages like data.table and Rcpp offer dramatic performance enhancements. Also, parallel computing is straightforward
8. Not Just for Statisticians:
   • R’s applications range from web development to machine learning (tidymodels, caret, mlr3) to reporting (Quarto, bookdown)

Closing Thought: Every tool has its strengths. The key is to understand and leverage them effectively.

Tidyverse

We’ll spend a lot more time on going over the details of the Tidyverse!

Intermezzo: A brief course History

General Course Outline

Tuesdays 08.00 - 12.00

• 08.00 - 08.30 Recap of key points from last weeks exercises
• 08.30 - 08.45 Introduction to theme of the day
• 08.45 - 09.00 Break
• 09.00 - 12.00 Exercises

About

Course Description

• Basically, what can you expect to learn and what do I expect that you learn: DTU Course Base

Course Resources

• Text Book: “R for Data Science 2e” by Hadley Wickham, Mine Çetinkaya-Rundel, and Garrett Grolemund
• Course site: https://r4bds.github.io

Course format

• Active Learning: Very strong emphasis on students working in groups, rather than me talking
• The focus is on you working actively, not me talking
• I will not go through all preparation materials in class
• Proper preparation is a prerequisite for completing lab exercises and maximising course yield
• I focus on supporting your independence, hence for some exercises you will have to seek out information (I’m not a good data scientist, I’m just slightly better at googling than others)

Exercise feedback

Weekly

• An exercise question will be highlighted
• Each group is responsible for crafting an answer to this highlighted question
• These answers will be hand-ins
• The following week, we will choose a random answer to be discussed in plenum
• Note: This starts from lab 2

Group Formation

• Modern Bio Data Science is a team sport!
• You have to form a group of 4-5 students with a Shared Bio Data Science / Bioinformatics Area of Interest
• You will work in these groups throughout the course
• You will do the final project in these groups
• You will attend the exam in these groups
• Group work is a very important meta skill for an engineer!
• Please fill in groups, see schedule for lab 1
• If you do not have a group, fill in your id and interest at an available group and someone might join you
• I aim to let you decide on the groups, I will of course be happy to help if needed

How to succeed in this course

• Prepare materials as instructed!
• Show up for class!
• Do the exercises!
• Do the project work!

Basically, show up, follow the curriculum and you will do fine!