How do I do protein structure homology-modelling similar to SWISS-MODEL without using a web api for computing and only use online databases for searching templates? I would like to use python (I heard salilab modeller can do this in some way) to integrate this into another additional script. It would be good to actually just search the templates using a web database and do the automodel on my computer. Is there any good way to do this role? Please suggest a similar way to do this if this has some difficulties. Thank you.
By the way, will it be efficient to use MODELLER or I-TASSER for building models with about 15~30 amino acid mutations for 200~450 residue proteins? I need to automate multiple tasks so time can be important if I-TASSER takes like hours to complete one. If one modelling task can be done within 35 minutes in a I7-6700K CPU and 32GB RAM computer, I will definitely do it in I-TASSER.
In my experience, modeling numerous proteins of all sizes, ITASSER is slow generally. I wasn't so concerned with speed, as much as I was with accuracy and throughput though, so I had roughly 300 jobs running each on a single core. If additional cores are available to the process, the job will no doubt be somewhat faster. It also depends on how close a published threading template is to your sequence. If the protein structure is very clear cut, the software will likely arrive at the answer quicker.
I've never used MODELLER personally, so have no idea how long it takes. As a general rule, protein structure simulation is slow though.
I'd hazard a guess that none of the modelling suites will give you a structure within an hour. I've simulated small peptides before, up to 50AA, which completed within an hour or two. Some of my simulations for larger proteins took weeks or even months to run though.