Darwin Help

Local A local alignment will be performed, this means that the best subsequences of seq1 and seq2 will be selected to be aligned. This type of alignment gives the highest possible similarity score of any alignment. This is sometimes called the Smith & Watermann algorithm.

Global A global alignment will be performed, this means that the entire seq1 is aligned against the entire seq2. This may result in a negative score if the sequences do not align very well. This is sometimes called the Needleman & Wunsch algorithm.

CFE A Cost-Free ends alignment is done. This is like a Global alignment, but deletions of one of the sequences at each of the end are not penalized. In some sense it is between a Local and a Global alignment.

Shake A forward-backward alignment is performed. This alignment iterates forward and backwards until the score cannot be increased. In its forward phase will start at the given positions for seq1 and seq2 and find the ends which give a maximal score. From this end, it will perform backwards dynamic programming to find the optimal beginning, and so on until convergence. This type of alignment is quite similar to a Local alignment, but can be directed to focus on a particular alignment, even though it may not be the best of the two sequences.

If the DayMat is omitted, the global variable DM (if assigned a DayMatrix) is used, else a PAM-250 matrix is constructed.

If in addition to the method, the keyword "NoSelf" is included, when sequences of peptides or nucleotides are aligned (excluding ProbSeq), self-matches are not allowed. That is, if a sequence is aligned to itself (being structurally the same string, this we call self-alignment), the self-match (which is trivial) will not be allowed. This is done by giving the alignment of a position with itself a large penalty. By doing this it is possible to find repeated patterns. I.e. an alignment with itself, where the identity is ruled out, will show any repeated patterns. In particular if the sequences align with an offset of k, then there is a k-long motif which is repeated in the sequence.

The method to find the approximate PamDistance and variance may not find the global maximum of the Score, it may find a local maximum. By using the argument "ApproxPAM=ppp", the search for the maximum will be started at PAM distance ppp. This may help when we know an approximation of the distance, or may provide a way of exploring the existence of other local maxima.