First Model: Template identification


Template identification
For template identification we simply choose the "Template identification" option from the Tools menu on the sidebar of the main page of the SwissModel server and then copy and paste your sequence into the field provided. In the case shown below I pasted the sequence from entry BCHD_RHOCB from Rhodobacter capsulatus, subunit BchD of magnesium chelatase (Mg-chelatase inserts Mg2+ into protoporphyrin IX, the first committed step in the chlorophyll biosynthesis pathway). Before submitting the sequence, pay some attention to the search parameters presented by the site: InterPro Domain Scan, Gapped Blast Query, Iterative Profile Blast and HHSearch template library search. The Gapped Blast Query is used for the identification of closely related homologues, while the Iterative Profile Blast is used for the identification of distantly related sequences. HHSearch template library search uses the method of Hidden Markov Model (HMM) based template library, again for identification of distantly related sequences. There is a detailed explanation in the help menu at the server. The run is not very fast, it usually takes some time, more than sufficient for a cup of coffee. Below is part of the output, which shows the few first hits found by the search:

BchD-template-search


The first of the hits is subunit BchI of magnesium chelatase, the others are either AAA+ ATPases or type A von Willebrand factor proteins. Note that only BchD has both these domains in a single polypeptide unit, none of the hits shown on the output has the two domains in a single polypeptide! This means that one will need at least two templates to make a homology model of the structure of this subunit. In addition, we can see that there is no known protein structure, which can be used to model the part of BchD located between these two domains. It is not unusual that for proteins, which contain several domains, not all domains can be modeled. However, modeling a couple of domains may still be sufficient, if these domains can answer some important functional questions.

It is also interesting to note that for the AAA+ domain, sequence identity between BchI and BchD is only 30%, very close to the threshold mentioned
previously:

BchI-BchD-template-alignment


Another way to analyze domain content of a protein is, as discussed earlier, by using the
InterPro database. As seen on the figure below, we again have two clearly identified domains, an AAA+ ATPase domain and a von Willebrand-type domain:

BchD-Interpro


Be aware that the template search server may also output homology models, if they are found in the PDB. In our case there is a model for the C-terminal domain of BchD, which was created when we made an electron microscopic reconstruction of the complex between BchI (the first subunit of Mg-chelatase) and BchD. This is not an experimental structure, and it may be difficult to know if a model found in PDB can be trusted. If you do not trust the model, you could make one of your own for comparison.
In the
next page we will start the actual modeling.

Homology Modeling