TIGR MIDAS Exercise (Wed Nov 22 2006) ===================================== Follow-up course of Microarray Data Analysis Nov 20-24 2006, PICB Shanghai Peter Serocka 1. Provide your raw expression data from yesterdays' exercise: Rename your annotated .mev output file from Spotfinder to manila-SLIDERNUMBER-YOURNAME.mev, where SLIDERNUMBER is either 1 (at Mark's row of desks), 2 (middle row), 3 (row at entrance door) and YOURNAME is Mark, John, etc. Copy this file to /picb/course/40-Manila-Raw-Data Note: For your convenience, your home directory -- which also contains your Desktop -- is accessible in all File-Open/Save windows as: /picb/course/User-Data/course01 (resp. course02 etc). 2. Help for using MIDAS software: A list of MIDAS functions and their icons is attached to this handout, as well a list of MIDAS output file names and their purposes. The full manual is provided as PDF: /picb/course/Documentation-TIGR/MIDAS_V2.19.pdf (searchable when viewed!) 3. Data quality filtering: a. In MIDAS, create a new project containing "Read single data file" and "Write process". Load your raw expression file from Step1 -- as parameter for "Read single file". As parameters for "Write process", choose Virtual Trim = Off and Output Trimmed Data = On. b. "Execute" the workflow. When asked to save the MIDAS workflow as a project, create a new folder named "filter1" in your homedirectory or your Desktop. As project filename, also choose filter1. All output files, including the "filter1.prj" file will be stored in your "filter1" folder. c. Switch to the MIDAS "Inspection" tab, and navigate to your project folder. View both .mev file (click right mouse button). Which one contains the good spots, which the bad? d. Repeat with different setting: In parameters for "Read single file", set Use ChannelA Flag = On, same for ChannelB. When executing the workflow store the project in a new folder ("filter2"), otherwise the .mev files in filter1 will be overwritten. Compare the filtered .mev files from both runs. What has changed? 4. Locfit (LOWESS) normalisation: b. Create a new (empty) project, and add "Read data directory", "LocFit Normalisation" and "Write process". Load some raw expression .mev files and execute workflow (save under new folder "norm1" and project name "norm1"). c. Find the output files in project folder "norm1" and its subfolders. (.mev data table and various plot, refer to list file types and names attached to this handout.) Compare RI plots of raw and normalised data. Does it make sense to normalise data of low quality? 5. Combining quality filtering and normalisation: a. Which are two ways to apply normalisation only to high-quality spots? b. Implement and execute one of them. (Flip page to No.6!) 6. Flip dye consistency checking a. Create a new project, later named "flip1", containing "Read flip dye" and "Flip dye consistency checking" and "Write process". Load one pair of dye-flipped files from /picb/course/31-Operon-ExprData-mev/ (Dr. Elumalai from Operon provided these in .gpr format -- converted to .mev with free TIGR ExpressConverter.) b. Make sure that Virtual Trim = Off and Output Trimmed Data = On for the "Write Process" function. c. Execute the workflow and examine the plots and the data tables. Have some inconsistent spots been trimmed out -- as seen from the .rrc plot? In which .mev output table can you find inconsistent spots?