Few generally available programs allow the flexible investigation of alternative alignments. A way of avoiding the problem is to make use of one of the earliest methods of comparing two sequences, the `dot-plot', or diagonal plot [Gibbs & McIntyre, 1970,Staden, 1982]. Although the dot-plot does not give an alignment directly, it is an effective technique to view simultaneously all similarities between two sequences. Similarities may be obvious in a dot-plot, but missed entirely by a dynamic programming sequence alignment program that only displays the top scoring alignment.
For two sequences A and B of length m and n respectively, a
matrix Dm,n is generated where each element Di,j represents
the similarity between sequence segments centred on Ai and
Bj. In its simplest form,
Di,j = 1 if
Ai = Bj and
Di,j = 0 if
.
A graph is plotted with A and
B on each axis and a dot plotted whenever
Di,j > 0. Regions of
sequence similarity appear as diagonal lines on the plot and repeats
as parallel lines. Insertions and deletions show up as steps in the
diagonals as the diagonal moves from one diagonal to another. The
most sophisticated dot-plot techniques calculate Di,j from a
sliding window, and score similarities by a pair-score matrix such as
BLOSUM62 or PAM250. Dotter [Sonnhammer & Durbin, 1995] provides a particularly rich
set of features including an interactive dot-plot `contrast'
adjustment that simplifies the interpretation of plots.