Protein Visualization

Visualization is provided by PV.
Model structure is coloured in a rainbow gradient (from N- (blue) to C-(red) terminus). It can be changed to B-factor colouring using a button below the sequence.
If more than one model has been calculated you can change the currently displayed one using buttons above the visualization window.

---

If a reference structure from RCSB PDB is available it is superposed and shown in a partially transparent gray. By default only the model structure is active, so can be clicked, and will have contacts displayed. Distance map used for the plot are changed along with the active model.
Currently inactive model can be hidden using a button above the visualization.
When changing the active model to the PDB reference structure, only the parts which have been modeled, (so appear in the Multiple Sequence Alignment used for DCA) are activated.

---

Clicking on the currently active model selects the residue under cursor in the visualization, on the plot, and on the sequence.

Direct Information Plot

Plot displays a number of top scoring, in terms of Direct Information, contacts, between residues more than three amino acids apart. Axes are annotated with residue indices, and automatically detected secondary structure elements are displayed along (alpha helices in magenta, beta sheets in cyan). Default plot shows the (protein length) top scoring residue pairs. Dots on the plot can be coloured according to the relative DI using a button below the sequence. Number of currently displayed points can be changed using the slider, or a textbox below it, or by modifying the "Lower bound for DI scores".
Below the slider, number of interactions shown is presented (including as a percentage), as well as the Positive Predictive Value (number of interactions with DI above the cutoff, with C-beta atoms closer than currently selected "Cutoff distance").

We have calculated Direct Information based on various models:

1. default: HH0 - sequences found using HHblits with e-value cutoff e⁰ (=1);
2. HH4 - sequences found using HHblits with e-value cutoff e^-4 (=0.0001);
3. JH0 - sequences found using JackHmmer with e-value cutoff e⁰ (=1);
4. JH4 - sequences found using JackHmmer with e-value cutoff e^-4 (=0.0001)

which can result in differing scores. Currently selected DI file can be changed with the buttons above the plot.

Selecting Selections can be made on the plot either using a traditional selection box (by clicking and dragging the mouse cursor on the plot), or by freehand drawing (by holding the Shift button while dragging).
Clicking on the plot clears current selection.
Holding the Ctrl button allows adding selections. To improve readability, consecutive selections are marked in a different colour.

Contact bonds All plotted interactions which fall within one of the selections will be added as lines between given residues in the protein visualization (as long as both residues are present in the structure). By default they are coloured according to the selection colour, but this can be change to either plot-point-based colouring (when "Colouring of the plot" is set to "Rainbow"), or distance-based ("Colouring of the contact bonds" set to "Distance").

Alternative plot modes By default only DI scores are shown, binary-coded - all interactions scoring above the given DI cut-off are displayed. Plot colouring can be changed to a rainbow gradient, according to DI, using "Colouring of the plot" button.

Other plot modes can be selected using buttons immediately below the sequence:

• Direct Information vs model contacts (side by side). This keeps the default plot in the lower triangle, while changing the upper triangle of the plot to a binary-coded distance map of the model (residues with C-betas closer than the specified cut-off distance are shown).

• Direct Information vs model contacts (overlayed). Gray points indicate pairs of residues of the model closer than the distance threshold, with low DI (False negatives). Red points are interactions with DI above the cut-off but further away in the structure (False positives). Blue points are interactions close in the structure, with high DI scores (True positives).

When reference structure is available:

• Model vs reference contact maps (overlayed). Purple points indicate pairs of residues which are in contact (distance between C-beta atoms lower than the specified cut-off) both in the model and in the reference structure. Red are contacts present only in the model, blue - those exclusive to the reference structure.

• Model vs reference distances (side-by-side). Lower triangle of the plot shows rainbow coloured map of pairwise distances for the model, upper triangle displayes the same for the reference structure. Upper bound of the distances used for colouring depends on the maximum value for the currently active model.

Sequence

Clicking on the sequence residues marks them also on the structure, and in the DI plot. To select consecutive residues, click and drag. To undo selection, click on the selected residue (or click and drag to deselect multiple residues).

Topology

For each model a full-chain topology is calculated - the most often appearing knot signature, when closing the protein chain in a 100 random directions (further details can be found in KnotProt). This is then visualized as a simplified knot icon, and corresponding numerical name.

If a refererence structure from the RCSB PDB is known, same analysis is performed for this structure, for comparison. Additionally, we give a basic interpretation of the compared results, taking into account the presence of gaps in the reference structure - which can influence topology calculations.

When changing the displayed model, both topology results and comparative interpretation (when available) change accordingly.