Homology modeling

What is homology modeling?

Homology modeling, also called comparative modeling, is a method for building a target protein structure from one or more related proteins with experimentally determined structures. It aligns a target sequence to selected templates and transfers their structural framework, so reliability depends on template choice, alignment, insertions, loops, and side-chain treatment.

A complete workflow searches and ranks templates, reviews a target–template alignment, constructs the model, and evaluates the result. Strong global sequence identity can still conceal errors around active sites, indels, domain boundaries, and long loops.

ProteinIQ does not run a dedicated comparative-model builder. This workflow begins with a model generated by a specialist service or local package, then connects repair, stereochemical validation, and reference comparison without claiming that downstream tools generated the model.

When to use homology modeling

• A structural template is defensible. Use it when a related protein has an experimentally determined structure and the alignment can be reviewed.
• Template assumptions need inspection. Compare sequence, coordinates, and local geometry before using a model downstream.
• Provenance can be retained. Keep templates, the alignment, model-building settings, and validation evidence with the exported model.

Benefits of homology modeling

• Template-derived assumptions are explicit. The selected structural evidence and its coverage can be inspected.
• Family-specific detail can be preserved. A suitable template can inform conserved structural regions.
• Residue-level comparison is possible. The resulting model can be compared with known structures.

Primary limitations

• The comparative model is external. ProteinIQ reviews rather than builds the model.
• Alignment errors propagate. Incorrect correspondence directly affects coordinates.
• Low-identity regions remain uncertain. Loops and insertions often need particular caution.

Homology-modeling decisions and interpretation

Template selection should consider assembly, domain coverage, ligands, mutations, resolution, and conformational state—not sequence identity alone. Inspect the alignment manually near functional regions.

Validation cannot prove a model correct, but it can identify implausible geometry and quantify similarity to a chosen reference. Retain the template identity, alignment, modeling settings, and every coordinate repair with the result.

How to run homology modeling online

1. Search for structurally characterized homologs. Select templates outside ProteinIQ.
2. Review the target–template alignment. Inspect gaps, domain junctions, and functional residues.
3. Build the comparative model externally. Use a dedicated service or package and retain its settings.
4. Review geometry. Import the PDB and run PDBFixer, MolProbity, and Ramachandran Plot.
5. Compare and export. Use US-align with a relevant reference and export provenance with validation results.

How homology modeling works

Import an external comparative model, repair common coordinate issues, inspect geometry, and compare it with a selected reference structure.

1. Select templates. Search for structurally characterized homologs and select templates outside ProteinIQ.
2. Review the alignment. Review the target–template alignment, especially gaps, domain junctions, and functional residues.
3. Build the model externally. Generate the comparative model with a dedicated homology-modeling service or package.
4. Check geometry and clashes. Import the PDB into the ProteinIQ workflow and run PDBFixer, MolProbity, and Ramachandran Plot.
5. Compare with structural evidence. Use US-align against a relevant reference and export the model with its provenance and validation results.

Inputs and outputs

Check formats before running, then inspect and download the result from every workflow step.

Inputs

Research input

FASTAPDBmmCIF

An externally generated homology model in PDB format and, when available, a relevant experimental reference structure.

Outputs

Prediction and review outputs

PDBmmCIFCSVJSON

A repaired model, MolProbity and Ramachandran geometry results, plus US-align structural comparison metrics.