oo-intro.Rnw

<<knitr, echo=FALSE>>=
opts_chunk$set(tidy.opts = 
               list(width.cutoff = 50, 
                    tidy = FALSE),
               fig.align = 'center',
               stop_on_error = 1L,
               comment = NA,
               prompt = TRUE)
options(width = 60)
@ 


\section{Object-oriented (OO) Programming}

% OOP basic concepts
\begin{frame}{Object-oriented Programming (OOP)}  
  \begin{block}{Object-oriented vs Procedural programming}
    
    \begin{itemize}
    \item OOP introduced in 1970s in Smalltalk but gained wider popularity in 1990s with programming languages like C++ and Delphi
      \bigskip
    \item Traditional (procedural) programming - data and functions decoupled
    \item Object-oriented programming - data and functions tied together in \textbf{objects}
    \end{itemize}
  \end{block}
  
  \begin{block}{OOP concepts}
    \begin{itemize}
    \item \textbf{Abstraction} - related data is stored and handled together
    \item \textbf{Inheritance} - code reuse by hierarchy of more-to-less general object types (classes)
    \item \textbf{Polymorphism} - the most appropriate function is called based on the dataset (e.g various \texttt{plot} functions)		
    \end{itemize}
  \end{block}
\end{frame}

% explain OOP modelling on a simple example (without emphasis on syntax)
\begin{frame}[fragile]{Procedural vs Object-oriented Programming}

  \begin{columns}[t]
    \begin{column}{0.38\textwidth}
      Procedural programming
      \begin{scriptsize}          
<<procedural-area,echo=TRUE, tidy = FALSE>>=
area <- 
  function(x1,y1,x2,y2){
    abs(x2-x1)*abs(y2-y1)
  }
area(0, 0, 5, 5)
@ 
\end{scriptsize}
    \end{column}

    \begin{column}{0.62\textwidth}
      Object-oriented programming
      \begin{scriptsize}
<<oop-area,echo=TRUE, tidy=FALSE>>=
setClass("Rectangle", 
         representation = representation(
           x1 = "numeric",
           y1 = "numeric",
           x2 = "numeric",
           y2 = "numeric")
         )
setGeneric("area", 
           function(obj) standardGeneric("area"))
setMethod("area", "Rectangle", function(obj) {
  abs(obj@x2 - obj@x1) * abs(obj@y2 - obj@y1)
})
rect = new("Rectangle", x1=0, y1=0, x2=5, y2=5)
area(rect)
@
\end{scriptsize}
    \end{column}
  \end{columns}

\end{frame}

% S3 and S4 in R
\begin{frame}{OOP in R (1)}
  \begin{block}{S3}    
    Older and less formal framework with no explicit class definitions. 
      Many parts of base \R use S3, e.g. plotting, linear modelling, ... 
      \begin{itemize}
      \item limited introspection, single inheritance, single dispatch, instance-based
      \end{itemize}
  \end{block}

  \begin{block}{S4}        
    Full-fledged object-oriented framework, de-facto standard for most modern packages and 
    required for Bioconductor packages.
    \begin{itemize}
    \item introspection, multiple inheritance, multiple dispatch (introduces a small overhead)
    \end{itemize}
  \end{block}
\end{frame}
  
\begin{frame}[t]{OOP in R (2)}
  \begin{block}{S4 Reference classes}
    Introduced in \R-2.12
    \begin{itemize}
    \item mutable objects, single inheritance, single dispatch, fields and methods in class definition, 
      methods associated with classes (rather than generics)
    \end{itemize}    
  \end{block}
\end{frame}

% Description of the working example
\begin{frame}[t]{Course working example}

  \begin{block}{Working example revisited}
    \small
    Working example for this course will be \textbf{manipulating DNA/RNA sequence data}.
    \smallskip

    Functions we would like to have:
    \begin{itemize}
    \item \code{readFasta()} - read in a single sequence from a FASTA file
    \item \code{id(), seq()} - return the ID of sequence and the sequence (accessors)
    \item \code{rev()} - return reverse DNA/RNA sequence
    \item \code{length()} - return DNA/RNA sequence length
    \item \code{comp()} - return complementary DNA/RNA sequence
    \item \code{transcribe()} - return RNA sequence for DNA sequence	
    \end{itemize}
  \end{block}

  \begin{block}{Goal}
    \small
    The final product should be an R package using S4 framework. But we need to get there, 
    so lets start with a procedural and S3 implementation...	
  \end{block}
\end{frame}