MULTIPLE ALIGNMENT ANALYSIS TOOLS
USER INSTRUCTIONS
This server will provide you with information about a possible
functional and/or evolutionary two-way split in a protein family. All three
programs take multiple sequence alignment of the family as the input.
ANCESTRAL CENTERS program will, for each position without a gap, return
a score. The smaller (i.e. more negative) the score, the more likely is
this positon to contribute towards the split. CLUSTERING will divide
your sequences into two groups with respect to the selected positions.
Given a division of the protein family into two groups, REVERSE CLUSTERING will rank
positions in the
alignment according
to their fitness with respect to this division.
All three programs take CLUSTAL alignments as input. If you have a
CLUSTAL-like alignment - such as T-Coffee - please replace "T-Coffee" (or
other)
with "CLUSTAL". You can also submit a "left-justified" alignment, with
names of the sequences submitted to the left hand window (names are
optional). Sequences should contain only amino acids from the standard alphabet,
with no "X". Please, replace "Xs" with some other letters, or, simply,
remove the sequences containing "X" from the alignment. For the
CLUSTERING program, you will have to select columns which will
be used to obtain the clustering. You should select at least 2 valid
columns (i.e. without a gap), and not more than 100. For REVERSE CLUSTERING, you have to
give a division of your family into two groups. Neither of the groups should be empty.
ANCESTRAL CENTERS will return a table, where, for each position without
the gap in the alignment, you get the hypothetical "ancestral amino
acid=one center", the score of this model, two hypothetical "centers",
the score of the two-center model, and the the difference of the
scores. The smaller the number in the last column, the more likely is
this position to be important for the split. CLUSTERING will
return a table, with scores of each of the sequences with respect to
two clusters, assignment to one of the clusters with respect to these
scores, and the absolute value of the difference of scores. REVERSE CLUSTERING will return
a table, containing a score for each position without a gap in the alignment. The larger the score,
the more likely is position significant for the proposed division.