Functions: Theory and Application (in R)

Athit Kao, PhD
UCI Bioinformatics Support Group - April 25, 2018

whoami

$Athit Kao$

R enthusiast
PhD, Biomedical Sciences, UCI
IGB Biomedical Informatics Training program alumnus
Focus on proteomics mass spectrometry and bioinformatics

Prerequisites

Understand content from previous deck: http://learn.athitkao.com/presentation_functions1.html
Basic programming experience necessary
R and RStudio installed
Consider a simple use case for programming in your field, e.g.:
- Filter specific data from file
- Munge proprietary data format into CSV
- Calculating RPKM for sequencing data

Theory

Readability, part II

Encapsulation via functions is an aspect of code legibility
Others include:
- Documentation (comments and READMEs)
- Formatting (indentation and line wrapping)
- Variable naming (obvious and consistent names)
There is a balance appropriate for you/your audience:

Generalized Concept of Functions

Function Argument Order

If argument names aren't specified (e.g. “first = first”), then they must be in order
https://www.safaribooksonline.com/library/view/r-in-a/9781449358204/ch09s05.html

Anonymous Functions (in lapply)

Function without a name that is evaluated without need of assigning it to a variable
A.k.a. lambda expression
https://lukesingham.com/anonymous-functions-in-r-python/

Multiple Arguments (in lapply)

First argument always the iterated variable
Must explicitly name additional arguments

The Assignment Operator

Use “<-” or “=”?
I use “=” for consistency (with other programming languages) with no issues
Consistency even within R

Environments and Variable Scoping in R

Scoping refers to the visibility of variables in different environments
Global: Can be referenced from anywhere
Local: Accessible only within its environment
http://adv-r.had.co.nz/Environments.html
R uses “lexical” (a.k.a. static) scoping rules

Applications

(in R)

Microsoft Excel

Pros

Quick editing and prototyping
WYSIWYG plots
Widely known and used program
Many similar alternatives (e.g. Sheets, Calc, Numbers, etc.)
Programming/automation with Visual Basic

Cons

Limit of 1M rows by 16K columns
32bit version has 2GB file limit
Manual editing error-prone
Complex visualizations not possible
Third-party packages close to non-existent (e.g. advanced statistics, machine learning, etc.)
Closed source

R Programming Language

Pros

Quick prototyping towards minimum viable product
Can generate complex interactive visualizations (plots, reports, etc.)
No practical limit on data size
Easily reproduce results and adapt code to changes
You can do almost anything purely in R

Cons

Steep learning curve
Third-party package developers not forced to follow a set standard
Open source

Split-Apply-Combine Paradigm

Multithreading Split-Apply-Combine