Download Installing and Using R and Bio Conductor - Lecture Notes | CS 5890 and more Study notes Computer Science in PDF only on Docsity!
CS 5890: Installing and Using R and BioConductor
For gene expression analysis, we can use a variety of tools in R. These notes are to help students install R on their own computer (or CD or flash drive), and then run a few simple commands.
Installing R
Run the executable file available at the following website:
http://cran.r-project.org/bin/windows/base/R-2.3.1-win32.exe
Follow the installation wizard that pops up. Note that you will probably choose to follow the English directions and accept the installation agreement. It’s probably best to select the default “User Installation” rather than customizing. If you want, you can put R on the start menu and create a desktop icon. This will all take just a few minutes.
Note that you can install R on a CD (probably best to use a rewritable CD with nothing else on it) or a flash drive. To do this, install R on a campus computer’s desktop (or somewhere else where you can get to the installed files), and don’t worry about creating a start menu folder or desktop icon. Once this is done, access the bin\Rgui.exe file in the created folder. You might want to create a shortcut from this file to somewhere else in the created folder you can easily access. Then copy the entire created folder to your CD (or flash drive). This will allow you to run R on any machine that can read your disk. Note that some older machines might not be able to read rewritable CD’s, but that shouldn’t be a problem in any student computer lab on campus.
Running R: A Simple Example
You can run R for Windows by either double-clicking on the desktop icon or the shortcut you created to access the bin\Rgui.exe file in the folder created during installation. This will open up an interactive environment where we will be using many features of R. Use the “Introduction to R” .pdf file available through manuals portion of the help menu in R as a good reference for the basic syntax of common commands in R. The search facility through CRAN (http://cran.r-project.org) is also very useful.
Consider the following simple example to look at a few commands in R. Suppose we have midterm and final exam scores for five students. Let x be the midterm score and y be the final score:
Student 1 2 3 4 5 Midterm 50 66 72 75 92 Final 65 65 89 79 97
To define these in R, use the following commands: id <- c(1,2,3,4,5) x <- c(50,66,72,75,92) y <- c(65,65,89,79,97)
This will create three “vectors” named id, x, and y. Think of a <- b as being an arrow assigning the right-hand side (b) value to the left-hand side name (a). Note that we can type these commands directly into R, or type them in a text editor (like Notepad) where we can save them for later use. If we have commands in a text file, we can either copy and paste them into R, or we can reference them using the source command (more on that later).
Suppose instead that the data were in some external file that we wanted to read in. For exam- ple, suppose the data are in the comma-delimited file found in a:\datafiles\scores.csv, with three columns labeled student, midterm, and final. Then the following commands would do the same as the commands given above: dataset <- read.table("a:\datafiles\scores.csv",sep=",",header=T) id <- dataset$student x <- dataset$midterm y <- dataset$final
Note that the $ is like a pointer to the variable inside the data set.
Look at a simple graphical representation of these data using a scatterplot by using the plot command: plot(x,y,main=’Scores’,xlab=’Midterm’,ylab=’Final’,pch=16) ?plot Note that the main, xlab, ylab, and pch are options to affect the appearance of the plot, and the? will give a pop-up window with a description of the plot command and its options. (As an aside, the graphical abilities of R are one of its great strengths. You can create very nice graphics in a variety of formats; type ?Devices for more.)
We could add points or lines to the plot by doing the following:
points(x=c(60,60),y=c(70,75),pch=1,col=’red’) abline(v=70,col=’blue’,lty=2)
Note that col and lty are options to affect the appearance of the plot. The v is for a vertical reference line; use h for horizontal.
We will use the square brackets [ ] to subset objects in R. For example, suppose we wanted to print out the final scores for all students who scored more than 70 on the midterm.
t <- x > 70
Then t is a vector of TRUE/FALSE
y[t]
Or, do it by hand:
y; y[c(3:5)]
To quit R, type q(), and for now, click ‘No’ when asked if you want to save the workspace image. (Equivalently, type q(’no’).) If you’ve done a lot of work that would take a long time to run again, and you will want to pick up where you left off in another R session, you can save the workspace image, either by saying ‘Yes’ to this option, or by an explicit command in R (either way, these .RData files can be quite large):
save(file="a:\worksinprogress\historyname.RData")
Then read this back in later using:
load(file="a:\worksinprogress\historyname.RData")
Bioconductor Packages
There are many “automatic” functions that come with R; however, most of what we’ll use in this class will require specialized functions dealing with gene expression data. These func- tions are included in separate “packages” that must be downloaded and installed.
Once you have installed R, you will need to get the packages necessary to do the analyses for this course. You can get them one at a time as needed during the course. (This is why you’d want to use a rewritable CD if you’re going that route for installing R.) For example, suppose you know you need the R package called affy (either because you know what it does or because you tried to do something in R and R stopped and told you it needed this package). There are many ways to get packages, such as installing them using the drop-down menus in R. However, you can use a pre-written function to get most packages:
Sys.putenv("http_proxy"="http://proxy.usu.edu:80/") Sys.getenv("http_proxy")
These first two lines may be necessary on-campus to access
online resources.
source("http://www.stat.usu.edu/~jrstevens/get.packages.R") need.pkgs <- c("affy") # or c("affy","vsn"), for example get.packages(pkgs=need.pkgs)
Note that the source command will go to the referenced file and run the code there in R. Running this the first time may take a minute or two because R will also go install other “dependent” packages.
Once a package has been downloaded and installed, it will only be used in an R session if you explicitly request that it be loaded (to save on memory space). For example, if we want to look at an image of a particular microarray, we would first need to load the affy package (using the library function) and then call the necessary functions. To see which CEL files are available in a directory:
library(affy) cels <- list.celfiles("C:\folder") cels
To read in particular CEL file(s):
data.0 <- ReadAffy(filenames="C:\folder\celfile.CEL") data.1 <- ReadAffy(filenames=c("C:\folder\cel1.CEL", "C:\folder\cel2.CEL")) data.2 <- ReadAffy(celfile.path="C:\folder") length(data.2)
This creates an AffyBatch object in R. Its length tells how many arrays are in it.
To select a specific array from the AffyBatch object:
array.1 <- data.1[,1]
To look at the array image (with a title as given):
image(array.1,main=’mouse array image’)
To look at gene names (probe set names):
gn <- geneNames(data.1) gn[1] # Name of first gene pn <- probeNames(data.1) pn[1:13] # Names of first 13 probes
But how does R know which genes (or probesets) are on this array? We need a “decoder ring.” Probe information is available in CDF files. R will automatically download and install the necessary packages when it encounters a CEL file. (Alternatively, you can download and install the packages yourself, by hand.) You can also create your own CDF file for custom arrays.
To look at intensities for specific probe set(s) and to put them together in one matrix:
pm.gn1 <- pm(data.1,gn[1]) mm.gn3 <- mm(data.1,gn[3]) mat <- cbind(pm.gn1,mm.gn3) colnames(mat) <- c(’pm1.gn1’,’pm2.gn1’,’mm1.gn3’,’mm2.gn3’) mat
A few graphical summaries of these data (four arrays here) provides motivation for normal- ization:
library(RColorBrewer)
May need first: get.packages(pkgs="RColorBrewer")
par(mfrow=c(2,2)) cols <- brewer.pal(4, "Set3") boxplot(data.2,col=cols,names=1:4) hist(data.2, col=cols, xlab="Log2 intensities", lwd=2,lty=c(1,1,2,1)) legend(12,0.2,1:4,col=cols,lwd=2,lty=c(1,1,2,1))
