Limits...
Evolutionary potentials: structure specific knowledge-based potentials exploiting the evolutionary record of sequence homologs.

Panjkovich A, Melo F, Marti-Renom MA - Genome Biol. (2008)

Bottom Line: We introduce a new type of knowledge-based potentials for protein structure prediction, called 'evolutionary potentials', which are derived using a single experimental protein structure and all three-dimensional models of its homologous sequences.The new potentials have been benchmarked against other knowledge-based potentials, resulting in a significant increase in accuracy for model assessment.In contrast to standard knowledge-based potentials, we propose that evolutionary potentials capture key determinants of thermodynamic stability and specific sequence constraints required for fast folding.

View Article: PubMed Central - HTML - PubMed

Affiliation: Departamento de Genética Molecular y Microbiología, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Alameda 340, Santiago, Chile.

ABSTRACT
We introduce a new type of knowledge-based potentials for protein structure prediction, called 'evolutionary potentials', which are derived using a single experimental protein structure and all three-dimensional models of its homologous sequences. The new potentials have been benchmarked against other knowledge-based potentials, resulting in a significant increase in accuracy for model assessment. In contrast to standard knowledge-based potentials, we propose that evolutionary potentials capture key determinants of thermodynamic stability and specific sequence constraints required for fast folding.

Show MeSH
EvP accuracy at different structure and sequence clustering cut-offs. (a) ROC curves for the different EvP sets depending on the structural clustering of the PDB space. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves. (b) ROC curves for the different EvP sets derived using different cut-offs of sequence identity in the MSA. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2643939&req=5

Figure 1: EvP accuracy at different structure and sequence clustering cut-offs. (a) ROC curves for the different EvP sets depending on the structural clustering of the PDB space. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves. (b) ROC curves for the different EvP sets derived using different cut-offs of sequence identity in the MSA. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves.

Mentions: The clustering of the structural space affects the selection and specificity of EvPs for model assessment. Various combinations of thresholds for structure similarity (that is, 80% and 90% of Cα atoms within 4 Å) and sequence identity (that is, 90%, 80%, 50%, 20%, and 10%) were applied to obtain 10 different sets of representative chains. EvPs calculated from the strictest clustering corresponding to 90% sequence and structural identity resulted in the most accurate assessment of the model accuracy as measured by their maximal accuracy (ACC), the area under the curve (AUC), false positive rate (FPR), and true positive rate (TPR) (that is, 99.5% AUC, 97.4% ACC, 2.3% FPR, and 97.0% TPR; Table 1). Variation of sequence identity for clustering had a marginal impact on the accuracy of the EvPs. However, a decrease of only 10% in the cut-off for the structural similarity had a larger impact, reducing the ACC and the TPR of the EvPs up to approximately 2% and 4%, respectively. Therefore, the accuracy of the EvPs for model assessment decreases as more structural variability is allowed within the clusters that represent the structure space (Table 1; Figure 1a; Table S1 in Additional data file 1).


Evolutionary potentials: structure specific knowledge-based potentials exploiting the evolutionary record of sequence homologs.

Panjkovich A, Melo F, Marti-Renom MA - Genome Biol. (2008)

EvP accuracy at different structure and sequence clustering cut-offs. (a) ROC curves for the different EvP sets depending on the structural clustering of the PDB space. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves. (b) ROC curves for the different EvP sets derived using different cut-offs of sequence identity in the MSA. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2643939&req=5

Figure 1: EvP accuracy at different structure and sequence clustering cut-offs. (a) ROC curves for the different EvP sets depending on the structural clustering of the PDB space. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves. (b) ROC curves for the different EvP sets derived using different cut-offs of sequence identity in the MSA. The inner panel zooms into the upper-left corner of the ROC curve to better show the differences between the curves.
Mentions: The clustering of the structural space affects the selection and specificity of EvPs for model assessment. Various combinations of thresholds for structure similarity (that is, 80% and 90% of Cα atoms within 4 Å) and sequence identity (that is, 90%, 80%, 50%, 20%, and 10%) were applied to obtain 10 different sets of representative chains. EvPs calculated from the strictest clustering corresponding to 90% sequence and structural identity resulted in the most accurate assessment of the model accuracy as measured by their maximal accuracy (ACC), the area under the curve (AUC), false positive rate (FPR), and true positive rate (TPR) (that is, 99.5% AUC, 97.4% ACC, 2.3% FPR, and 97.0% TPR; Table 1). Variation of sequence identity for clustering had a marginal impact on the accuracy of the EvPs. However, a decrease of only 10% in the cut-off for the structural similarity had a larger impact, reducing the ACC and the TPR of the EvPs up to approximately 2% and 4%, respectively. Therefore, the accuracy of the EvPs for model assessment decreases as more structural variability is allowed within the clusters that represent the structure space (Table 1; Figure 1a; Table S1 in Additional data file 1).

Bottom Line: We introduce a new type of knowledge-based potentials for protein structure prediction, called 'evolutionary potentials', which are derived using a single experimental protein structure and all three-dimensional models of its homologous sequences.The new potentials have been benchmarked against other knowledge-based potentials, resulting in a significant increase in accuracy for model assessment.In contrast to standard knowledge-based potentials, we propose that evolutionary potentials capture key determinants of thermodynamic stability and specific sequence constraints required for fast folding.

View Article: PubMed Central - HTML - PubMed

Affiliation: Departamento de Genética Molecular y Microbiología, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Alameda 340, Santiago, Chile.

ABSTRACT
We introduce a new type of knowledge-based potentials for protein structure prediction, called 'evolutionary potentials', which are derived using a single experimental protein structure and all three-dimensional models of its homologous sequences. The new potentials have been benchmarked against other knowledge-based potentials, resulting in a significant increase in accuracy for model assessment. In contrast to standard knowledge-based potentials, we propose that evolutionary potentials capture key determinants of thermodynamic stability and specific sequence constraints required for fast folding.

Show MeSH