Venkataraghavan Ragunathan Sorry I did not understand you.

That depends very much on the system you want to analyze - For example, in Docking you would choose different programs depending on whether you want to dock two proteins, a highly flexible peptide to a protein, or a fairly rigid organic molecule to a protein. You would have to decide whether you want to do global docking, or whether you have prior knowledge of the approximate binding site and want to sample only a small fraction of the surface of the protein, e.g. the binding pocket of a natural ligand. It also depends on the magnitude of the task: For virtual screening of 100'000nds of compounds, you might choose a relatively fast algorithm that uses some simplifications to speed up the process, To then re-screen the best hits with a more sophisticated method. There is not one generally applicable best solution - but some are better for the specific task at hand than others.

In multiple sequence alignments, some of the parameters determine how the program handles alignment gaps - parameters you frequently can set are gap insertion and gap elongation penalties which determines whether your programs tries to align large compact sequence blocks or breaks them up by small gaps to match as many amino acids or nucleotides as it can - The best values for these parameters depends on the diversity of your sequences.