Scansite Frequently Asked Questions (FAQ)

How do I interpret the Scansite score for my site?

Scores in Scansite start at 0.000 if the sequence optimally matches the given motif, and the scores increase for sequences as they diverge from the optimal match. The score histograms and the percentile are the main ways to tell how good a score is for a given motif. If your site ranks in the best 0.2% of all sites, that is quite good (this is the "high stringency" threshold value we use for most of the motifs). The Z score will also tell you how many standard deviations from the mean it is (the Z score will be negative because lower Scansite scores are better). For database searches, the percentile is calculated from all proteins you searched (a whole database, for example). For the Motif Scan program, the percentile is calculated from a reference database search, which is the vertebrate section of the Swiss-Prot database.

It may be worthwhile to consider the surface accessibility of the predicted site, as shown by the plot below the graph of sites found. A site that can actually be reached by an interacting protein is better than a buried one. However, our surface accessibility is calculated from the sequence, and not with reference to any known structures, so its accuracy may be limited.

Another way to assess the validity of a site found is to check whether it is conserved in related organisms. Clicking on the sequence will show you the position of the site within the protein, and on that page is a convenient link to submit the site to BLAST.

How do I make a matrix to use in Scansite?

To use your own binding motif in a database search, you will need to define it in a text file and import the text file into Scansite. The format is a matrix of typically 20 columns and 15 rows. The columns are scores for the 20 amino acids at each position, and the 15 rows are positions of each residue in the binding pocket. Scansite requires one residue to be invariant in the motif sequence, such as a tyrosine for motifs recognized by tyrosine-kinases. (For motifs that recognize more than one residue in this position, such as kinases that phosphorylate both serine and threonine, multiple residues can be treated as equivalent in the fixed position.) The middle position, row 8, holds the fixed residue in the matrix. Rows 1 through 7 hold the scores for positions to the left (N-terminal side) of the fixed position, and rows 9 through 15 hold the scores for those to the right (C-terminal) side. Finally, at the top of this 20 by 15 matrix should be a row of column headings indicating the one-letter amino acid code.

The actual number of columns in the matrix can be higher or lower than 20. If your peptide library screen did not include some residues, such as cysteine or tryptophan, these columns can be left out of your matrix and Scansite will assign them default values of 1 everywhere except in the fixed position, where the default score is 0. Alternatively, if your motif targets the C terminus (as PDZ domains do), you can include an extra column giving scores for the C terminus, labeled with the symbol "*" (asterisk). The N terminus can be given scores in a column labeled "$" (dollar sign). (When absent, Scansite assigns the N and C terminus columns values of zero by default.) Selenoprotein researchers can add a column of scores labeled "U" for selenocysteine (the U symbol is present in recent releases of the Genpept database). When this column is absent, Scansite scores selenocysteines as cysteines. Finally, columns can be included for the degenerate symbols "B" (aspartate/asparagine), "Z" (glutamate/glutamine), and "X" (any residue) to score sequences containing these symbols in the public databases, although there is probably no research reason to include these in an input matrix. As a result of all these options, an input matrix can contain as many as 26 columns, or even only 1 column naming the fixed residue (such a matrix would not be very informative, however). The order of the residue columns is not important ("A" does not have to be first, and so on).

An example will make this more clear. Below is a sample matrix. The full matrix might have 21 columns in this case (20 residues plus the C terminus), but only seven are shown here in order to fit well on the page. The first line in the text file contains the column headers (residue letters and the asterisk). These should be separated by tabs. In practice, it is easiest to do this by creating the whole matrix in a spreadsheet program and saving it as a tab-delimited text file. The next 15 lines are the 15 positions relative to the fixed center at row 8. This motif is specific for tyrosine (or phosphotyrosine) in the center position, so tyrosine is the fixed residue. It must be given the value 21 in row 8 for Scansite to recognize it as the fixed residue. All other values in row 8 must be zero. (For multiple residue possibilities at the fixed position, each of them must have the value 21 and the remaining residues must all be zero.) This motif also requires the C terminus to be four residues away from the center tyrosine (so the column "*" is given the value 21 at that position). In all the other rows, scores are given to represent the relative binding affinity for each residue type at each position in the sequence.
ACDE...WY*
1111...110
1111...110
1.630.299.089.4...2.913.670
8.880.830.283.92...5.931.450
1.565.816.493.31...5.357.070
0.115.5715.00.03...0.314.950
6.790.388.22.17...3.370.590
0000...0210
5.123.615.531.91...8.637.030
2.184.535.526.11...8.940.710
0.634.075.342.51...4.493.630
4.745.91.563.1...2.396.820
0000...0021
1111...110
1111...110

The scoring system ranges from 0 to roughly 21. Giving an individual amino acid a score of 1 at one position in the motif indicates that no preference exists, positive or negative, for that particular amino acid in that position. Giving all amino acids in one position of the motif a score of 1 (i.e. making all values in a single row of the matrix equal to one, as in the first two rows in the table shown above) indicates no preference exists for any particular residue type at that position in the motif. A score of 15 means that residue has 15 chances out of 21 to be the selected residue, which is a very strong affinity. It is not necessary to normalize the scores at one position to add up to 21; Scansite will normalize them during program execution. Values higher than 21 are permitted if you wish to indicate very strong affinities. Values between 0 and 1 can be used for very low affinities, but negative numbers cannot. Beware that the scoring function uses natural logarithms, so values less than 1, particularly those less than 0.5, strongly penalize for that particular residue in a motif. In fact, the penalty of negative selection from a matrix value of 0.1 in the final score is equivalent (though opposite) to the positive selection obtained with a value of 10 for another residue in the motif. Consequently, choose values less than 1 (other than 0) carefully. This matrix favors the C terminus, but for motifs not showing any special affinity for the N or C terminus, scores in the "$" and "*" columns (if present) should be zero. No commas or other punctuation should separate matrix values. Only tabs are permitted between columns.

While the center position must have a fixed residue, it is also possible to fix other positions as well (as done in this example for the C terminus). As another example, suppose your motif requires aspartate in row 5 (three residues N-terminal to the fixed center). To do this, give that aspartate the value 21 in row 5, and make all other amino acid values equal to 0. Conversely, you should avoid giving residues the special score of 21 if they are not intended to be fixed positions; use values like 20.9 or 21.1 instead.

Scansite Home