Structural Class Probabilities
In the structural class probability plot, we see that the alpha superclass has a probability of about 0.4, the irregular superclass has a probability above 0.3, alpha-beta superclass probability is slightly above 0.2, and the beta superclass has a probability near zero. This means that psa-request is confident that the protein sequence has properties that tend to rule out the all-beta superclass. Furthermore, the alpha superclass is slightly more probable than the other two candidates, but not by much.
Looking at the macroclass probabilities, we see that the apb (antiparallel bundle) macroclass is more probable than any other macroclass, but that ir (general irregular) runs a close second.
Secondary-Structure Probabilities
This plot provides a detailed view of secondary-structure probabilities. Each row corresponds to a different secondary structural state, and each column corresponds to a different residue position. The probabilities of each residue being in each of the structural states are depicted using contour lines of constant probability in increments of 0.1. Areas surrounded by many contour lines are regions of high probability, while areas outside of the contours have low probabilities of less than 0.1.
The structural states are shown in four groups, with horizontal lines drawn for each row that denotes a structural state, and empty rows in between the groups. The first group has the three beta strand states:
buried
STRAND
exposed
Some strands appear in the model as amphipathic, with alternating
exposed (to the solvent) and buried (not exposed to the solvent) states.
[The other strand state on the middle line is called "average strand";
that is probably for strands of "undeclared" exposure, using "average"
statistics, but I'm not certain of this. Looking at this plot, it would
appear to be the sum of the other two rows. -- rgr, 27-Sep-00.]
The four turn states are numbered 1 through 4 from the bottom up (with the "2" on the "TURN" line implicit). These denote a tight turn in the secondary structure, and usually appear in the Type-1 models as part of a beta hairpin structure. Turn states are comparatively rare in the Type-1 models overall, so their probabilities are usually low. However, they are shown in order to aid researchers who have other reasons to suspect that their sequence includes a turn, i.e. evidence of a hairpin. See the "Summary of Type-1 DSMs" table for more information on the typical number of turns included in each of the Type-1 models.
Next, the helix-buried and helix-exposed states are displayed together in the same way as for strands, except that there is no helix state for "undeclared" exposure. The advantage to showing the helix-exposed and helix-buried states separately in this format is that amphipathic helices show up clearly as an easily-recognizable pattern of high probability that alternates between these two states, with a clear periodicity of 3.6 residues per helical winding. In this example, the sequence is predicted to be mostly amphipathic helix, but there are two breaks in the pattern, so it looks like three helices.
Finally, the loop state appears across the bottom. There is only one type of loop as far as the DSMs are concerned, so this "group" uses only one row. The loop state is defined loosely as "none of the above," i.e. if it's neither helix nor strand nor turn, it must be a loop. As a consequence of this definition, the probabilities in each row of the transcript add up to one. Since the Type-1 DSMs were constructed to reflect the fact that proteins are roughly half secondary structure and half loop, the loop probability is usually above ten percent, and often above 50 percent, especially near the start and end of the sequence.
For example, the 40th residue has probability between 0.7 and 0.8 of being in a loop, because there are seven contour lines surrounding the point on the loop row for this residue. The actual value is 0.709, as can be gleaned from the full transcript.
Looking at the plot as a whole, three amphipathic helices are clearly depicted with their buried and exposed residues (there might be four if one considers the middle helix to consist of two helices with minimal transition between them). In contrast, the strand state is improbable: every point along the strand row has no more than two contour lines around it.
The probability contours in this plot, together with the reasonable hypothesis that the protein belongs to the alpha parallel-bundle class (which has four helices of approximately equal length), support the following summary prediction of the protein's secondary structural topology:
(Loop)-(amph. helix)-(short loop)-(amph. helix)-(short loop or turn)- (amph. helix)-(short loop)-(amph. helix)-(short loop)
For reference, here are the DSSP secondary structure assignments for PDB locus 2mhr:
SS start end Helix 1 19 37 Helix 2 41 64 Helix 3 70 85 Helix 4 93 109
The three graphs in this plot show the probabilities for each residue
position being in a strand, turn, or helix. This is the same
information as in the transcript section of
the cover letter, which provides exact values for all residues, and
a subset of the information in the
secondary-structure probabilities plot, which also breaks the helix
and strand probabilities down by exposure.
For example, the 20th residue has a probability of greater than 0.9
of being in a helix, and negligible (< 0.02) of being in a turn or
strand. The remaining probability, about 0.03, is the probability of
being in a loop state. (The exact values for all residues are included
in the cover letter.) The undulating
helical probabilities in the third graph support the presence of four
helices of nearly equal lengths connected to each other by short loops
or turns.
Go to:
Please direct your questions and comments about these Web pages and
the PSA e-mail server to:
Strand/Turn/Helix Probabilities
Bob Rogers
<rogers@darwin.bu.edu>
Last modified: Mon Mar 12 13:29:55 EST 2001
BioMolecular Engineering Research
Center
Boston University, Boston Massachusetts