|
Motif Discovery Assessment: Participant Instructions |
|
 CSE Home |
About Us |
Search |
Contact Info |
|
|
This paragraph describes the instructions given to participants. Each
participant ran his or her program on each data set, selected the
single best motif, and reported the location and content of each of its
occurrences, and the parameter settings used to find them.
Participants were free to mask repeats in the input sequences or vary
the parameters for different data sets or postprocess the output to
eliminate low-complexity motifs, as they saw fit. We requested that
participants describe the exact procedure used, which would be helpful
for later users.
Unlike the CASP competition for protein structure prediction, it is clearly easy to cheat in this assessment: by having your program consult the TRANSFAC database, for instance, you could probably improve your program's performance considerably. This was intended to be an assessment among de novo predictors, and participants were instructed not to modify their programs to consult such databases of known binding sites or otherwise employ methods that would not be available in a real application. Our collective goal was to educate ourselves and our users about which types of programs work best for various types of real applications, where the "correct answer" is not known. Format for submitting resultsResults were submitted in two text file, a Result File and a Parameter File. The formats for these two files are given below.
|
 |
Computer Science & Engineering University of Washington Box 352350 Seattle, WA 98195-2350 (206) 543-1695 voice, (206) 543-2969 FAX [comments to tompa] |
