Hello,

I have been using Modeller a few times for simple approaches but, currently, I have to face a more complex problem. I have the crystal structure of my template (740 residues, PDB: 4J0M) and I want to model three putative homolog sequences coming from a different organism. In that sense, the putative protein sequences are 1100 residues long, so I can only model the region that it locally matches with my template. For this reason, I have to use a local alignment (Smith-Waterman algorithm) between the target and the template and you may imagine which kind of problems I have to face when trying to run Modeller with this alignment: as it is a local alignment, there are multiple positions in the pdb file that have been skipped in the alignment so I have to remove those atoms from the pdb file one by one. Additionally, there are missing positions in the crystal template so I have to remake the alignment (deleting from the template fasta file these residues) each time I find missing atoms in the pdb, plus removing the atoms in the pdb file that do not match with the local alignment.

I am pretty sure there is a more intelligent and practical solution to this problem since I have to do this for more than 700 residues and it can take hours for only one model.

Since I am a new Modeller user and I have only used it for simple purposes, any help you can provide it would be very appreciated.

Thank you very much for your attention and sorry for the inconvenience.