Alignment scores with fixed gap penalties:
------------------------------------------
1003 686 361 905 92 59
686 1002 359 718 68 32
361 359 938 368 68 21
905 718 368 1012 68 52
92 68 68 68 939 90
59 32 21 52 90 963
Alignment scores with gap opening and extension:
------------------------------------------------
1003 686 390 905 133 95
686 1002 388 718 105 82
390 388 938 391 133 73
905 718 391 1012 118 79
133 105 133 118 939 116
95 82 73 79 116 963
The reason why the first two alignments are so similar, while the latter two are so different, can be found in either alignment score matrix above. Note that 1MBC and 1YMC score very highly against one another (905), suggesting a high degree of similarity between the two sequences. Since the two sequences are so similar, it is unlikely that we will need mutliple insertions and deletions to successfully establish a high degree of identity between them. Therefore, a smaller the gap extension penalty is unlikely to make gaps more attractive to the alignment of 1MBC and 1YMC.
On the other hand, the alignment score for 1LHS vs. 1LH1 is relatively low (32), suggesting that the sequences are not very similar. We are likely to need multiple insertions and deletions in order locate regions where the two sequences are somewhat similar, and improve the overall alignment. The gap opening and extension method allows us to both introduce more gaps into the alignment and make use of the fact that gaps tend to cluster together, by using a large gap opening penalty and a much less sharp gap extension penalty. The result is an alignment with long contiguous gap regions, and a larger overall score.