9. Moving from sequence to atomic structure scenario

In this section we are going to obtain the initial model of our target sequence, in this case the Hgb \alpha subunit. To perform this task we are going to use the Scipion protocol chimerax-model from template. Although this protocol offers several different possibilities to get the right result (see Appendix Model from template for details and use cases), we are going to consider that we already have a template, that we have found in the previous step and that will be used as protocol input. In addition, although the protocol also allows to get a model including two chains modeled simultaneously using the same template (multichain modeling), in our example we are going to model only one chain, the Hgb \alpha subunit. Fig. 9.1 shows the workflow that we are going to detail in this section. Other possibilities of the protocol usage will be suggested at the end of the chapter.

Fig. 9.1 Scipion framework detailing the workflow to generate the first model of the human Hgb \alpha subunit.

9.1. Downloading the atomic structure

Once identified the template that we are going to use as structural skeleton of our sequence, we import it into Scipion with the protocol import atomic structure (see Fig. 9.2 (1) and Appendix Import atomic structure). Select the option for importing the atomic structure from ID (2), write the PDB accession code (3) and execute the protocol (4).

Fig. 9.2 Importing the atomic structure 1PBX.

You can visualize the imported structure (5) in ChimeraX (Fig. 9.3). By selecting chain A in the upper menu (1) you can distinguish the Hgb \alpha subunit (2).

Fig. 9.3 Atomic structure 1PBX visualized with ChimeraX. Hgb \alpha subunit (selected chain ) is shown green-highlighted.

9.2. Structural models of human metHgb subunits from templates

Modeller [Sali and Blundell, 1993] is one of the computational web services used by ChimeraX, which provides the interface to run the program. Working with Modeller requires a license key, which is provided free of charge for academic users. Modeller allows two types of modeling computations to generate theoretical models, template-based (sequence homology) and template-free (de novo, only for missing segments). In this tutorial we are going to consider the first one: structure prediction by sequence homology. Requirements for this type of modeling are the template structure and a sequence alignment including sequences of target and template.

Important

before starting!! We are going to use a ChimeraX-derived Scipion protocol for the first time in this tutorial (chimerax-model from template, Appendix Model from template). Remark that this use of ChimeraX is completely different from the use of ChimeraX as a visualization tool, as we have seen previously. By using the ChimeraX graphics window, opening it from the Scipion button Analyze Results we can observe protocol results but we CANNOT save anything in Scipion. However, using ChimeraX as a tool, as it is the case in Scipion ChimeraX-derived protocols, we can perform different tasks, taking advantage of the available ChimeraX tools and, finally, we CAN save the obtained results and the working session in Scipion.

• Preparing your sequence alignment:

  In addition to the ways to obtain the sequence alignment using ChimeraX, this alignment can be also generated in the Scipion protocol chimerax-model from template (Appendix Model from template). This protocol allows selecting between pairwise and multiple sequence alignments. Besides producing more reliable alignments, especially for more distantly related sequences, multiple sequence alignments (MSA) provide more structural information than pairwise alignments; they locate conserved regions in the molecule, thus improving predictions of structural arrangements due to mutant residues or residues that differ between template and target sequences [Pearson, 2013]. For this reason, in this tutorial we are going to perform a MSA. Additionally, you can also test the available tools to perform pairwise alignments.
  Besides target and template sequences, other sequences are needed to accomplish a MSA. The type and number of the sequences included depends on the sequence conservation, although they have to allow differentiating conserved regions. As an example, our MSA will include four more Hgb \alpha subunit sequences from organisms located between human and fish in the evolutionary scale: Equus caballus (Horse), Oryctolagus cuniculus (Rabbit), Meleagris gallopavo (Wild turkey), Aldabrachelys gigantea (Aldabra giant tortoise). Download these sequences one by one from UniProtKB database filling in the import sequence protocol form with the appropriate accession codes, P01958, P01948, P01948, and P83134, respectively (Fig. 9.4). A similar process has to be followed for Hgb \beta subunit, importing UniProtKB sequences P02062 (HBB HORSE), P02057(HBB RABIT), G1U9Q8 (G1U9Q8 MELGA) and P83133 (HBB ALDGI).
  

  Fig. 9.4 Importing additional sequences to perform the MSA.

• Access to Modeller in ChimeraX:

  The protocol chimerax-model from template allows direct opening of the MSA in ChimeraX and then, access to Modeller via web service. Fill in the protocol form (Fig. 9.5 (1)), including the template 1PBX previously imported (2), the particular chain of interest (use the wizard to select it (3)) and the target sequence of human Hgb \alpha subunit (4). Since we plan to improve the alignment by including additional sequences to align (5), they will have to be added next (6). Finally, select one of the multiple sequence alignment tools (7).
  

  Fig. 9.5 Importing the multiple sequence alignment in ChimeraX.

  ChimeraX will be opened including this time the MSA together with the ChimeraX graphics window (Fig. 9.6). The template selected chain is shown green-highlighted in both windows. As you may observe in the alignment, Hgb \alpha subunit is a quite conserved macromolecule; there is only one gap in the alignment because PRO (Proline) 47 residue disappeared throughout the evolutionary process.

  

  Fig. 9.6 Opening the multiple sequence alignment in ChimeraX.

To complete the form that will allow us to get some atomic models of the target sequence in Modeller web service, we have two possibilities: a) to select Tools (Fig. 9.6, red arrow) -> Sequence -> Modeller Comparative , or b) clicking with the right mouse inside the Seqview box (Fig. 9.6, green arrow) and selecting Structure -> Modeller Comparative in the pop up window. A new window of Modeller Comparative will be opened (Fig. 9.7 (A)), that we have to fill in. Sequence alignments (1) should include the template sequence. In the Target sequences section (2) we should include the target sequence that we would like to model, HBA_HUMAN_P69905 in this particular case. Modeller license key has to be included here (3). The number of output models, 5 by default, can be also specified (4). Since the target sequence that we would like to model should include non-water HETATM residues (HEME group) we are going to select this choice as Advanced option (5). Finally, press OK (6) to start the computation of potential models for your target sequence.

Fig. 9.7 (A) Completing the form to access to homology modeling with Modeller. (B) Resulting model scores. (C) ChimeraX Models panel.

In ChimeraX main graphics window, lower left corner, you may see the status of your job. After a while, five possible atomic structures, from now ahead models, are retrieved for the target sequence (Fig. 9.7 (B)) together with their assessment scores. Column GA341 of Modeller Results indicates the score derived from statistical potentials (values in [0, 1]; > 0,7 for reliable models). Column zDOPE (normalized Discrete Optimized Protein Energy) score depends on the atomic distance (negative values for the better models). You can check every model number in ChimeraX’s main menu (Tools -> Models (C)).

For this tutorial we are going to select model #3.2 (Fig. 9.7 (B), green arrow). Renaming this model is the first step to save it. We can rename the model by typing in the ChimeraX command line:

rename #3.2 id #4

The renamed atomic structure will appear in the Models panel (Fig. 9.7 (C), green arrow). To track this new atomic structure in the Scipion workflow, we can write in the ChimeraX command line:

scipionwrite #4 prefix model_from_modeller_3_2_

In case that the Advanced option “Include non-water HETATM residues from template” (Fig. 9.7 (A, 5)) didn’t include the HEME group in the retrieved models, an alternative option to have the model with the HEME group (residue 144 from the atomic structure #2 chain A) could be:

rename #3.2 id #4
save /tmp/chainA.cif format mmcif models #4
open /tmp/chainA.cif
select #2/A:144
save /tmp/HEME.cif format mmcif models #2 selectedOnly true
open /tmp/HEME.cif
scipioncombine #4,5

scipionwrite #6 prefix Hgb_alpha_

Note

We have saved the HEME group of the template chain A in a new file that will be opened as model #5. Finally, the combination of models #4 (retrieved aminoacid model of the target sequence) and #5 (HEME group of the template chain A) generates a new model #6 that will be saved in Scipion. A different model ID could be selected by the user adding to the last command line modelid n.

After closing ChimeraX, you can visualize (Fig. 9.5 (8)) your full predicted model (Fig. 9.8) that includes the HEME group (1). The string that we have included as prefix in the command line scipionwrite will allow us to follow the atomic structure in a more simple manner. You can check the prefix in the name of the saved atomic structure (Hgb_alpha_Atom__struct_6_006815) in the Models panel of Fig. 9.8 (1). Interestingly, the suffix number of the saved atomic structure (006815) stands for the ID protocol number.

Fig. 9.8 Initial model of human metHgb \alpha subunit, including the HEME group (blue).

In a similar process, you can also obtain the initial atomic structure of the human metHgb \beta subunit. Take into account that in this last case the HEME group is the residue 148 of the chain B. The command lines are similar to the previous case of the metHgb \alpha subunit if you also select the model #3.2.

rename #3.2 id #4
save /tmp/chainB.cif format mmcif models #4
open /tmp/chainB.cif
select #2/B:148
save /tmp/HEME_B.cif format mmcif models #2 selectedOnly true
open /tmp/HEME_B.cif
scipioncombine #4,5
setattr #6/A c chain_id B
scipionwrite #6 prefix Hgb_beta_

In addition, we have included a command to change the chain id of the second polypeptide from A to B. In general, to change the chain ID you have to write:

setattr #model_number/old_ID c chain_id new_ID
setattr #model_number/old_ID r chain_id new_ID

Note that in the new version of ChimeraX you have to remove c from setattr 6/A c chain id B:

setattr #model_number/old_ID chain_id new_ID

9.2.1. Additional exercises for practising

Since the protocol chimerax-model from template allows to use other options, inspect by your own the possible result obtained by:

1. Using as input only the target sequence of the human metHgb \alpha subunit.

2. Using as input the same atomic structure template and the target sequences of both the human metHgb \alpha and \beta subunits. Improve the alignment of the human metHgb \alpha subunit with additional sequences and improve the alignment of the human metHgb \beta subunit with your own sequence alignment that contains about 30 sequences.

9.3. Option of recovering the ChimeraX session

If for any reason you decide to go back and check a different model from the five models initially provided by Modeller, you can do it by using chimerax-restore session protocol (Appendix ChimeraX restore session). This protocol may be used whenever session had been saved, specifically after using protocols chimerax-rigid fit, chimerax-model from template, and chimerax-map subtraction. In addition to the ChimeraX command line scipionss, command lines scipionwrite and scipioncombine also save ChimeraX session by default. So, if you want to restore a previous session just open the form (Fig. 9.9, 1), and include the session that you’d like to restore (2).

Fig. 9.9 Restoring session in ChimeraX.