Fig 1.
The figure illustrates the process of submitting a job to the cluster via JMS.
Fig 2.
Flow chart depicting the modeling algorithm incorporated by PRIMO.
The steps involved in modeling using PRIMO can be seen as an interactive process in which the user can supply and edit input as they see fit, while PRIMO chains these steps together to model protein targets of interest.
Fig 3.
Workflow followed to test the PRIMO backend scripts.
A) Overview of the steps followed when modeling known protein targets from the PDB. B) Filtering steps involved to reach the final 5,869 targets.
Fig 4.
The diagram shows how PRIMO can be separated into frontend web interface, which facilitates communication between the users and the backend scripts saved in the JMS, which perform the steps involved in homology modeling, which are numbered as 1) template identification, 2) template-target sequence alignment and 3) model building and evaluation.
Fig 5.
As with other stages, the progress bar (A) and job history set (B) are displayed on the page. These allow navigation to within the current jobs and between different modeling jobs, respectively. Completed jobs are shown in green, those awaiting user input in yellow and running jobs in purple. The list of models are tabulated (C), ranked by their DOPE Z-scores. This table can be used to select and download models produced, link to their evaluation page, as well as show them in the interactive protein viewer (D). In the viewer is the top model (teal), zoomed in to show ADP and Mg ions that have also been modeled from the template.
Fig 6.
DOPE Z-score results obtained from testing the backend scripts of PRIMO.
Modeling sets are divided into their respective bins and alignment programs used, as shown in the key. The ‘Remodel’ column depicts the scores for the targets modeled using themselves as templates. The ‘Target’ and ‘Template’ columns show the DOPE Z-scores of the known structures of the targets to be modeled and the templates used for modeling in each bin.
Fig 7.
Assessment of the backend scripts of PRIMO.
Results shown are the average RMSD values (A) TM scores (B), GDT-HA scores (C) and lDDT scores (D) of models produced for each sequence alignment program in each bin.
Fig 8.
Assessment of model refinement options.
DOPE Z-score results (A) and B) RMSD values (B) obtained when testing refinement options provided by MODELLER. Models are divided into sequence identity bins as in Fig 7. Results are shown for refinement options very slow, fast and none.