To use the benchmark to evaluate the performance of your fold-recognition method you need to search the benchmark's library of 301 folds with each of the 68 probe sequences of the benchmark.
For each probe sequence, your method should output a list of the 301
folds sorted in decreasing order of compatibility.
For example, for the probe sequence 1cpcl the output should be of the form:
1cpcl 1cpca 56.10
1cpcl 1cola 24.29
1cpcl 1mbc 18.82
1cpcl 1hbg 17.76
1cpcl 1dsba 16.01
etc.
where the third (optional) column indicates the compatibility score
of the pair.
When you have obtained all the 68 output files, merge them into a single file (of 68 times 301 lines) and e-mail it to me (fischer@mbi.ucla.edu).
You will receive the sensitivity and selectivity scores of your output be e-mail.
If you wish to be included in the table of methods evaluated so far, you can request to do so by e-mail. It is your own responsibility to avoid tuning your method on the benchmark's sequences. The UCLA-DOE benchmark www page is not responsible for any inaccuracies in the results received. However, when possible, some kind of verification will be carried out by us before posting your results in the table. If your method is not available for verification, this will be indicated in the table.