Bugs in USC

Help
2009-09-02
2013-05-20
  • Benjamin Nota

    Benjamin Nota - 2009-09-02

    Hi,

    I am trying to use USC classification in TIGR MEV 4.0 (and tried 4.2 and 4.4 as well), but there are some bugs.

    I use a pentium 4, windows XP machine.

    The first bug is when I save my results after training, the created txt file can't be used for a next classification (with the option classify from file). This text file contains in the first two columns my gene names, but I saw in the manual that the second column should contain numbering (1,2,3,...). If I replace the second column with 1,2,3,... in excel it does work. However, I don't know if the results are still correct then or can be trusted?

    The second bug I noticed, is when I check the classification results, a gene list (classifier) is created. This gene list is different than the genes used for the heatmap results! I would like to save these genes and their expression values (from the heatmap), to do clustering, etc., but the genes from the result heatmap are not the ones from the classifier list.

    I hope some-one can fix these bugs, because I really like to use this analysis!

    Thank you!

    Best regards, Ben

     
    • Eleanor Howe

      Eleanor Howe - 2009-09-04

      Hi Ben,

      Well, you've found a really good bug here. Unfortunately we here at DFCI didn't write the USC module, so we're still looking into what may be wrong here. I've emailed our collaborators at University of Washington, where USC was written, to see if they can shed some light on the problem.

      We are fairly sure that your workaround for the first bug is fine. It seems like a file-parsing issue that you have addressed with Excel. As for the second bug, though, I just don't know what's going on. I will post again if I hear from our collaborators. Otherwise, we'll put this bug on our list and see if we can work it out.

      I hope this was of some help. Please let us know if you find out anything, ok?

      Eleanor

       

Log in to post a comment.