Tutorial: Examining Genome3D Structures

This tutorial will introduce you to viewing and comparing Genome3D's 3D structural models.

Example 1: EIF6, "B(2)GCN homolog"

Let's start with a sequence that, though it doesn't have a structure itself, is highly similar to sequences that do (PHYRE found around 75% sequence identity). The Genome3D models for this structure are highly similar to each other.

  1. Open  P56537 in a new browser tab.
  2. Look at the Predicted Domains: all predict domains of similar superfamilies on the same part of the sequence (by mousing over the domain regions, we can see they've all got confidences of 99% or 100%). Let's look at the structures that Genome3D resources have predicted for this sequence...
  3. Under Predicted 3D Structures, click the grey domain region to the left of DomSerf. The domain will be highlighted with an orange glow to indicate it has been selected.
  4. Click View selected 3D Structures to see the selected structure.
  5. This will trigger the server to dynamically render an image of the DomSerf model for you (please be patient as loads may be high during this tutorial)
  6. We can superpose multiple structures by simply selecting overlapping grey domain regions and waiting for the image to update
  7. Let's have a proper look at the structures - select all the Predicted 3D Structures and then click PyMOL (please be patient; you may need to accept a dialogue window about how to open the file; if PyMOL isn't working on your machine please ask someone for help)
  8. Once the PyMOL structures have loaded, rotate them using the left mouse button and zoom in/out using the right mouse button. In this case the structures are so similar that the colours of the different models form a patchwork on the surface.
  9. Let's use ribbons to find tiny differences between the models: to the right of the all at the top-right of the PyMOL window, click S (ie Show) -> as -> ribbon
  10. Now zoom close in (right mouse button) and try to find tiny variations between models
  11. Try hiding/showing individual models by clicking on their identifiers, below all at the top-right of the PyMOL window.
  12. Close down the PyMOL window and this example's browser tab.

Example 2: UBE2QL1, "Ubiquitin-conjugating enzyme E2Q-like protein 1"

Let's look at a slightly trickier example: a sequence with less sequence similarity to any known structures (PHYRE found around 52% sequence identity). These Genome3D models are bit less similar than the ones in the first example.

  1. Open  A1L167 in a new browser tab.
  2. As in example 1, select all the Predicted 3D Structures and click View selected 3D Structures.
  3. To the left, under Display Options, click Colour by rainbow. This colours all the structures along the sequence from blue to red (N to C), which can help show the topology of the structure and can also help to highlight the similarities and differences between the models.
  4. Click PyMOL to fire up PyMOL.
  5. Have a good look around the structures. Try to trace the topology of the models from the blue end to the red end. Try to find areas in which the models are structurally similar or divergent.
  6. Let's use ribbons to help see these structural similarities/differences more clearly: to the right of the all at the top-right of the PyMOL window, click S (ie Show) -> as -> ribbon (or type show_as ribbon).
  7. Now let's add lines that connect equivalent residues in these models of the query sequence:
    • Click alignment at the top-right of the PyMOL window to activate it
    • To the right of the alignment, click S (ie Show) -> dashes (or type show dashes, alignment)
  8. Check that these lines help to show the regions of structural similarity (with short lines) and divergence (with long lines).
  9. To see how this relates to the sequence, click the S very close to the bottom-right of the PyMOL window. This will show the sequences of these models at the top of the PyMOL window. (Note: these sequences happen to be aligned because the models all start at the same residue but this is not the case for all superpositions - if you know how to specify your own alignment for PyMOL to use here, please let us know)
  10. Residues can easily be highlighted by clicking them on the structures or on the sequences. Try selecting a few residues on the sequence.
  11. Once you've selected some residues, show the side chains for these residues: to the right of (sele) at the top-right of the PyMOL window, click S (ie Show) -> sticks (or type show sticks, sele)
  12. You may also find it useful to click the down arrow symbol between the S and F very close to the bottom right of the window. This makes the structures automatically rock back and forth which can help you to keep resolving them as 3D structures.
  13. Close down the PyMOL window and this example's browser tab.

Example 3: OR5AL1, "Olfactory receptor OR11-184"

Let's look at a sequence that has less sequence similarity again to any known structures (PHYRE found around 20% sequence identity). Modelling at this level of sequence similarity can be a lot more challenging.

  1. Open  P0C617 in a new browser tab.
  2. As in the previous example, select all the Predicted 3D Structures, click View selected 3D Structures, select Colour by rainbow and click PyMOL
  3. Wait a moment for the server to process these requests and for PyMOL to startup with the superposition
  4. Take a good look around (left mouse button to rotate; right mouse button to zoom in/out)
  5. In this example, it is easier to find differences between the models. Nevertheless, even at this low level of sequence similarity to any known structures, the models for this sequence all have clear structural similarities.
  6. Let's show the backbone to investigate these models further: type show_as lines, (name n,ca,c)
  7. Now add the residue numbers and look around residue 263:
    • Type label n. CA, resi to show the residue numbers
    • Type orient i. 263 to zoom in on residue 263 in the models
  8. If you have any trouble with the above, click here to get a pre-prepared PyMOL script for this point in the tutorial.
  9. Rotate the structures whilst looking closely at the residue numbers. It may help to hide/show individual structures here by clicking on their identifiers, below all at the top-right of the PyMOL window.
  10. Notice that the residue numbers all mis-match on this helix. So, though this region is structurally similar in the different models, the resources have modelled equivalent structural locations for different regions of the sequence.
  11. Another way to show this phenomenon is to add lines between equivalent residues (as in the previous example):
    • Ensure that you have all structures selected at the top-right of the PyMOL window
    • To the right of the all at the top-right of the PyMOL window, click L (ie Label) -> clear (or type label) to clear the residue labels
    • Click alignment at the top-right of the PyMOL window to activate it
    • To the right of the alignment, click S (ie Show) -> dashes (or type show dashes, alignment)
  12. Zoom out slightly (right mouse button) and rotate around this helix (left mouse button). These lines connect the equivalent residues in different models. These lines cover large distances up and down the helix, which shows us that equivalent residues in the sequence have been modelled in different locations on the structure.
  13. Let's see how widespread this phenomenon is. Click Orient on the other PyMOL window or type orient) to reorient to viewing all the structures again.
  14. Now notice that there are other helices, both near the start and end of the sequence, for which these lines are very short indeed.
  15. In summary:
    • Though the sequence has low sequence similarity, all resources have generated models with clear structural similarity
    • Behind the overall structural similarity, there are some disagreements on which part of the sequence belongs to which part of the structure.
    • For some regions, both near the start and end of the sequence, all groups are in complete agreement.
  16. Close down the PyMOL window and this example's browser tab.

Thank you for completing this tutorial.