Related Experiment Videos
All are not equal: a benchmark of different homology modeling programs
1Stockholm Bioinformatics Center, Albanova University Center, Stockholm University, Stockholm, Sweden. bjorn@sbc.su.se
Protein Science : a Publication of the Protein Society
|April 21, 2005
Summary
This study benchmarks six protein homology modeling programs. Modeller, nest, and SegMod/ENCAD demonstrated superior performance, though none matched specialized side-chain modeling tools.
Area of Science:
- Structural biology
- Computational biology
- Biophysics
Background:
- Homology modeling is crucial for protein structure prediction.
- Alignment and template selection are key, but program performance is less understood.
- Previous benchmarks focused on alignment, not model construction programs.
Purpose of the Study:
- To benchmark six homology modeling programs: Modeller, SegMod/ENCAD, SWISS-MODEL, 3D-JIGSAW, nest, and Builder.
- To evaluate program performance based on physiochemical correctness and structural similarity.
- To identify leading homology modeling software for 3D model construction.
Main Methods:
- Comparative analysis of six distinct homology modeling programs.
- Evaluation metrics included physiochemical correctness and structural accuracy.
- Benchmarking against known correct protein structures.
Main Results:
- No single program excelled in all tested aspects.
- Modeller, nest, and SegMod/ENCAD consistently outperformed other tested programs.
- The older SegMod/ENCAD program showed competitive performance despite its age.
Conclusions:
- Modeller, nest, and SegMod/ENCAD are recommended for homology modeling.
- Current homology modeling programs have limitations in side-chain modeling compared to specialized tools.
- There is significant scope for improving side-chain modeling accuracy in general homology modeling software.