VIEWS: 2 PAGES: 34 POSTED ON: 2/21/2013
Computing for Research I Spring 2011 Lecture 1: January 5 Primary Instructor: Elizabeth Garrett-Mayer Introduction • Description: Students learn to use the primary statistical software packages for data manipulation and analysis, including (but not limited to): R, R Bioconductor, SAS, SAS macro, and Stata. Additionally, students will learn: how to use the division's high speed cluster-computing environment, how to practice the principles of reproducible research using Sweave in R, and how to use LaTeX and BibTeX for manuscript and presentation development. This is a three credit course. • Course Organization: This course is given by the entire division. Instructors will take turns giving lectures in their areas of expertise. • Textbooks: No textbook. Reading material (primarily found on the web) will be provided as necessary. • Prerequisites: Biometry 700 Evaluation • Grading: Instructors will give short exercises to be completed and turned into the primary instructor by the Wednesday of the week following when it was assigned (e.g., assignments given on Monday Feb 28 and Wednesday Mar 2 are both due on Wednesday Mar 9). Each assignment will count equally towards 75% of the course grade. There will be a final project which will account for the remaining 20% of the course grade. The remaining 5% of the course grade will reflect class participation. • Homeworks Policy: Homeworks are due by 5pm on the due date. All homeworks should be emailed to the primary instructor (garrettm@musc.edu) or turned in at lecture time. Asking for extensions on homeworks is strongly discouraged. However, it is expected that, on occasion, extenuating circumstances may arise. Therefore, the policy is that each student may request an extension on homework twice and the extension is to be no more than 2 days. After using two extensions, no more extensions will be granted except with a medical note. Contact Primary Elizabeth Garrett-Mayer Instructor: Website: http://people.musc.edu/~elg26/teaching/statcomputing.2010/statcomputingI.2011_files/statcomputingI.2011.htm Contact Info: Hollings Cancer Center, Rm. 118G garrettm@musc.edu (preferred mode of contact is email) 792-7764 Time: Mondays and Wednesdays, 2:30-4:00 Location: Cannon 301, Room 305V Office Hours: TBA, or by appointment Office Hours: The primary instructor will have office hours and be available by appointment. However, given the nature of the course, the primary instructor may not be knowledgeable regarding all of the topics covered. As a result, additional help may be needed to complete assignments from the lecturers. Be considerate and responsible in scheduling time with course instructors and recognize that they all have busy schedules. Course Objectives Upon successful completion of the course, the student will be able to • Import, perform simple analyses and produce graphical displays in Stata, SAS and R • Create new functions or commands in each of R, Stata and SAS • Generate professional quality scientific manuscripts and presentations using Latex along with statistical software • Perform standard power and sample size calculations using available software and simulations. • Operate the division’s cluster computer with batch computing Schedule, briefly • SAS • STATA • R • Batch processing • Latex + Sweave • Data management • Etc: power calculations, other packages Detailed Schedule Date Lecturer Topic W Jan 5 E. Garrett-Mayer Introduction; Overview and Principles M Jan 10 Jody Ciolino SAS: introduction W Jan 12 Sharon Yeatts SAS: IML W Jan 19 Renee Martin SAS: macros M Jan 24 Valerie Durkalski SAS: proc tabulate and proc report W Jan 26 Nate Baker SAS: Gplot M Jan 31 Annie Simpson SAS: ODS W Feb 2 Jordan Elm SAS: array processing M Feb 7 E. Garrett-Mayer STATA: introduction, “immediate” commands W Feb 9 Joan Cunningham STATA: data organization, manipulation M Feb 14 E. Garrett-Mayer STATA: exploratory data analysis; graphical displays W Feb 16 E. Garrett-Mayer STATA regression commands M Feb 21 E. Garrett-Mayer STATA: programming and do files W Feb 23 E. Garrett-Mayer R: introduction to object-oriented programming Detailed Schedule Date Lecturer Topic W Feb 23 E. Garrett-Mayer R: introduction to object-oriented programming M Feb 28 Caitlyn Ellerbe R: downloading packages/libraries; data input & output W Mar 2 Anthony Parker R: basic language structure (ifelse, where, looping) M Mar 7 Cody Chiuzan R: graphics W Mar 9 E. Garrett-Mayer R: exploratory data analysis M Mar 21 Yanqui Weng R: simulations; random number generation; sampling from distributions W Mar 23 Stacia DeStantis R: regression commands M Mar 28 Bethany Wolf R: bioconductor W Mar 30 Adrian Nida Batch processing (using R) and cluster computing M Apr 4 Mulugeta Gebregziabher Latex and Bibtex: manuscript production W Apr 6 Dipankar Bandyopadhay Latex and Bibtex: presentations M Apr 11 Betsy Hill Reproducible Research: Sweave and StatWeave W Apr 13 Amy Wahlquist Data management: RedCap M Apr 18 Annie Simpson Data management principles & Excel W Apr 20 Other packages (TBA) M Apr 25 Paul Nietert Sample size calculation software packages W Apr 27 Housekeeping • We are meeting in a regular classroom • Bringing laptops is encouraged • Data, code, etc. needed for class will be on the website prior to class • For optimal interface, install packages ASAP – R (http://cran.r-project.org/) – Stata (DBE helpdesk request) – SAS (DBE helpdesk request) – WinEdt (http://www.winedt.com/) • Create a bookmark to the course website Lecture Notes • Every lecturer will have his/her own style • Notes may be – prepared ahead of time and posted – Prepared and posted after the lecture – Nonexistent • Lecture notes will NOT be printed by the instructors prior to lecture. • If they are available and you would like a paper copy, it is your responsibility to print them out. Introduction • 2011: to be a successful biostatistician/epidemiologist, you MUST be competent on the computer. • Historically: students learned in labs from students • Moving forward: – many options for analysis and generation of results – Efficiency in computing is essential. – Your computer IS your lab! Data analysis software • In this course: – SAS – Stata –R • Many other options: SPSS S, Splus Epi Info GraphPad JMP Matlab JAGS Systat Minitab EGRET BMDP MedCalc Mathematica WinBugs GLIM …. SAS: History • SAS was conceived by Anthony J. Barr in 1966. As a North Carolina State University graduate student from 1962 to 1964, Barr had created an analysis of variance modeling language. From 1966 to 1968, Barr developed the fundamental structure and language of SAS. • In January 1968, Barr and James Goodnight collaborated, integrating new multiple regression and analysis of variance routines developed by Goodnight into Barr's framework. • By 1971, SAS was gaining popularity within the academic community. One strength of the system was analyzing experiments with missing data, which was useful to the pharmaceutical and agricultural industries, among others. • In 1976, SAS Institute, Inc. was incorporated. • The latest version, SAS version 9.2, was released in March 2008 SAS: functioning • SAS consists of a number of components, which organizations separately license and install as required. • Licenses expire! Software cannot be used after expiration (unless renewed) Why (or why not) SAS? • Most commonly used in pharma (although that may be changing!) • FDA likes SAS • Many jobs for MS statisticians and/or epidemiologists require SAS expertise • The most common language • Becoming less the choice of academia – Updates are less frequent than freeware – ‘pros’ of competitors are starting to outweigh the ‘pros of SAS • Licensing costs • Slow to add new functionality • Lack of consistency with syntax • Learning curve is slower than other programs that now have similar capability Stata • Stata is a general-purpose statistical software package created in 1985 by StataCorp. • Most of its users work in research, especially in the fields of economics, sociology, political science, biomedicine and epidemiology. • Relatively simple to learn yet powerful • Latest version is Stata 11 (released 2009). • Lots of add-ons for epi users Why (or why not) Stata? • Relatively inexpensive (especially as student or single- user) • Biomedical focus so output, functions are tailored to medical research • Fast and big: can handle and manipulate large datasets • Sophisticated with wide range of tools • Easy to learn language with consistent syntax • Graphics are not as good as other packages (although that has improved) • Programming (simulations, loops, etc) is more challenging R: History • R is a programming language and software environment for statistical computing and graphics. • The R language has become a de facto standard among statisticians for the development of statistical software, and is widely used for statistical software development and data analysis. • R is an implementation of the S programming language. S was created by John Chambers while at Bell Labs. R was created by Ross Ihaka and Robert Gentleman, and is now developed by the R Development Core Team. R is named partly after the first names of the first two R authors, and partly as a play on the name of S. • R source code is freely available under the GNU General Public License. • The capabilities of R are extended through user-submitted packages, which allow specialized statistical techniques, graphical devices, as well as import/export capabilities to many external data formats. A core set of packages are included with the installation of R, with more than 2460 (as of July 2010) available at the Comprehensive R Archive Network (CRAN). R: functionality • Freeware: latest version can be installed anywhere at anytime • Packages (aka libraries) that are user- contribute allow additional features/commands • Relatively simple interface Why (or why not) R? • Great for programming and simulations • Handles looping well • Flexible language • FREE! • User-contributes packages included in real-time (i.e., no delay in their availability) • Most PhD Biostatistics programs teach their students R and many/most academic statisticians in top programs use R. • Interfaces nicely with other programs such as Latex (Sweave), WinBugs, C, Emacs. • Can be clunky for data management. • Memory is not as good as SAS and Stata • Quality-control on user-contributed packages not evident Overview • Not a question of which one. • Question is “for my current problem, which package makes the most sense to use?” • Each has strengths and weaknesses Latex and Sweave • LaTeX is a document markup language and document preparation system for the TeX typesetting program. • The term LaTeX refers only to the language in which documents are written, not to the editor used to write those documents. In order to create a document in LaTeX, a .tex file must be created using some form of text editor. (e.g. WinEdt) • LaTeX is most widely used by mathematicians, scientists, engineers, philosophers, lawyers, linguists, economists, researchers, and other scholars in academia. • LaTeX is used because of the high quality of typesetting achievable by TeX. The typesetting system offers extensive facilities for automating most aspects of typesetting and desktop publishing, including numbering and cross-referencing, tables and figures, page layout and bibliographies. Latex and Sweave • Sweave is a function in R that enables integration of R code into LaTeX documents. The purpose is "to create dynamic reports, which can be updated automatically if data or analysis change". • The data analysis is performed at the moment of writing the report, or more exactly, at the moment of compiling the Sweave code with Sweave (i.e., essentially with R) and subsequently with LaTeX. This can facilitate the creation of up-to-date reports for the author. • Because the Sweave files together with any external R files that might be sourced from them and the data files contain all the information necessary to trace back all steps of the data analyses, • Sweave also has the potential to make research more transparent and reproducible to others. However, this is only the case to the extent that the author makes the data and the R and Sweave code available. Sample size and power • We don’t really use textbook formulas anymore to do simple power calculations (just like we don’t really invert matrices by hand when we analyze data). • There are a number of packages that quickly and easily perform simple power calculations • R, SAS and Stata can do some. • But, packages like Nquery, EAST and PASS do a lot more. • In some non-standard settings, simulations are required to determine power. Data management • Analysis of clean data is easy! • The real world: you will get messy data most of the time from your colleagues • Data management tools will help you; – Deal with messy data – Set up data capture approaches for your colleagues to minimize messiness • Excel, RedCap and general principles of data management for statistical analysis will be covered Example Patient # cycle # total ceramide levels S1P levels C18 ceramide S1P/C18 1 0 743.6 197.2 9.8 20.122449 3 625.6 177.9 9.9 17.969697 2 0 534.8 148.4 9 16.4888889 CR 3 461.6 182.8 10.8 16.9259259 5 527.3 151.4 11.5 13.1652174 3 0 760.5 214.5 12 17.875 4 0 359 167.3 4.3 38.9069767 3 375.9 125.3 4.6 27.2391304 5 475.6 116.2 4.4 26.4090909 5 0 394.1 163.1 5.7 28.6140351 6 0 848.7 132.5 10.8 12.2685185 3 1083.6 203.9 13.5 15.1037037 7 0 684.6 191.4 8.1 23.6296296 8 0 822.7 219.5 8.9 24.6629213 9 0 486.3 198 5.7 34.7368421 CR 581.3 186.8 9.6 19.4583333 699.6 42.3 11.4 3.71052632 561.7 130.4 6.7 19.4626866 754 320.6 14.4 22.2638889 Before getting started… • Types of files involved in statistical computing – Data files – Results files – Command/batch files – Function files – Graphics files – + more(?) • TIPS: – develop a common nomenclature for naming files and folders – Organize projects within folders Organization is key! • DO NOT overwrite old files (especially data files) • Save with a new name – Mousedata.xls (file sent from colleague) – Mousedata.clean.xls (your clean version of the data) • Use a consistent approach, but think ahead – Naming files *.new.* is not a good idea. You may have a new ‘new’ next week – Numerics are good, but if you think you may need more than 9 versions, consider how data2 and data10 would be alphabetized. Examples • For each Principal Investigator I work with, I have a folder • With the PI folder, for each project, I have a folder • For each time I get a new dataset (or work on a new grant) for that project, I have a folder named with month and year • Example: I:\\MUSC Oncology\\Kraft, Andrew\\VelcadeTrial\\May2008 I:\\MUSC Oncology\\Kraft, Andrew\\R01 June 2007 Examples • Within each folder of data analysis or grant development calculations, I use the same naming conventions for files: – Rbatch.R: a set of R commands that implement all of the computation or analyses – Rfunctions.R: a set of R functions that are used by the batch file – I always save the original data file from the investigator before making any changes – I add ‘clean’ to the datafile name and save it as a . Csv before use. – My Rbatch.R files always include a line sourcing in the data, including the folder where the data resides. Friends in Statistical Computing 1. Google is your friend 2. ‘Help’ functions and ‘see also’ links are your friends 3. ‘examples’ are your friends 4. Your fellow students are your friends Friends help friends figure out statistical computing! Using your noggin • Example 1: – SPSS is not included in this curriculum. – Can you not use it? NO! – Will you be able to learn it better and faster after having taken this course? YES! • Example 2: – We will probably not cover the R package nnc (Neareset Neighbor Autocovariates) – Does that mean you need to find someone to teach it to you? NO! – Will you be able to teach it to yourself? YES! • Example 3: – None of your instructors are computer scientists (except maybe Annie Simpson) – Does this mean that they are not qualified to teach you? NO! – Most of them are self-taught with regards to these techniques Final Thoughts for Today • THIS COURSE WILL POINT YOU IN THE RIGHT DIRECTION AND PROVIDE A SET OF TOOLS • IT IS YOUR JOB TO MAKE THEM FIT TOGETHER AND USE THEM AS A LAUNCHING PAD TO SOLVE PROBLEMS • Next up: Jody Ciolino with an intro to SAS! References • Background info on R, SAS, Stata, Latex and Sweave was all pilfered from Wikipedia.