Before doing a secondary structure prediction, click on the sequence of interest. This sequence becomes hilighted in yellow, and the "Choose Sequence" button becomes active. Click on the button to select the sequence.
When a sequence is selected, all columns in the alignment where that sequence has a gap are deleted. The consensus sequence of the profile is also set to that particular sequence, so the amino acids in the saved file are the sequence you're interested in. The latter step has no effect on the secondary structure prediction, since only the amino acid frequencies at each position (not the consensus sequence itself) are used in making the prediction.