Workshop: Protein structure modeling in Modeller, CollabFold, SwissModel
Description
- We will obtain protein structure based on templates
- Duration: 60 minutes
- Objectives: be able to
- Prepare protein sequences for Modeller
- Select a template based on sequence identity comparison
- Align target sequence with the template
- Model building
- Model evaluation
- CollabFold usage
- Swiss-Model usage
Jupyter notebook
AlphaFold2
Swiss-Model
Required software and resources
- Access to a Jupyter notebook evironment with Python 3, modeller, MDanalysis, nglview, Bio libraries
- Access to Google collab with GPUs
- To install modeller run in your conda environment “conda install -c salilab modeller”, use license key “MODELIRANJE”
Learning resources
Assignments
Modeller (based on tutorial)
- Build model for lactate dehydrogenase from Trichomonas vaginalis (TvLDH) based on a single template. Perfom loop refine by LoopModel, DopeLoopModel.
- Compare plots of energy score (DOPE) for model, template, model with auto loop refine, model with Dope loop refine.
Select YOUR PROTEIN sequence for modeling structure
- Find protein of your interest, BLAST it with “NON-REDUNDANT PROTEIN SEQUENCES” database and choose a sequence with 60% identity or less.
- BLAST the chosen sequence with “PROTEIN DATA BANK PROTEINS” database to check if it contains the 3D structure for your sequence (it must NOT have 3D structure in PDB).
AlphaFold2
- Predict 3D structure of YOUR PROTEIN (choose small one or fragment); use one of ChimeraX or Collab. Add sequence coverage plot and structure image.
Swiss-Model
- Predict 3D structure of YOUR PROTEIN. Compare with AlphaFold2 structure. Do NOT use AlphaFold structures as a template, only experimental ones!
(*) Repeat the Modeller task with YOUR PROTEIN. Compare obtained structure with AlphaFold2 and Swiss-Model ones. Do NOT use AlphaFold structures as a template, only experimental ones!
Troubleshooting
- Consult with the seminar protocol/recording
- Ask questions in Telegram