Limits...
Binding ligand prediction for proteins using partial matching of local surface patches.

Sael L, Kihara D - Int J Mol Sci (2010)

Bottom Line: The surface patches are encoded with the 3D Zernike descriptors.Unlike the existing methods which compare global characteristics of the protein fold or the global pocket shape, the local surface patch method can find functional similarity between non-homologous proteins and binding pockets for flexible ligand molecules.The proposed method improves prediction results over global pocket shape-based method which was previously developed by our group.

View Article: PubMed Central - PubMed

Affiliation: Department of Computer Science, Purdue University, West Lafayette, IN 47907, USA; E-Mail: lee399@cs.purdue.edu.

ABSTRACT
Functional elucidation of uncharacterized protein structures is an important task in bioinformatics. We report our new approach for structure-based function prediction which captures local surface features of ligand binding pockets. Function of proteins, specifically, binding ligands of proteins, can be predicted by finding similar local surface regions of known proteins. To enable partial comparison of binding sites in proteins, a weighted bipartite matching algorithm is used to match pairs of surface patches. The surface patches are encoded with the 3D Zernike descriptors. Unlike the existing methods which compare global characteristics of the protein fold or the global pocket shape, the local surface patch method can find functional similarity between non-homologous proteins and binding pockets for flexible ligand molecules. The proposed method improves prediction results over global pocket shape-based method which was previously developed by our group.

Show MeSH

Related in: MedlinePlus

Prediction performance using shape and pocket size information. ROC AUC values of pockets that bind to (A) ATP, FAD, FMN, NAD, and STR; (B) AMP, GLC, HEM, and PO4. (C) Average ROC AUC values over all ligand types. Top-3 prediction success rate of (D) ATP, FAD, FMN, NAD, and STR; (E) AMP, GLC, HEM, and PO4. (F) Average Top-3 success rate over all ligand types. *NT in x-axis denotes experiments with no threshold used.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC3100846&req=5

f4-ijms-11-05009: Prediction performance using shape and pocket size information. ROC AUC values of pockets that bind to (A) ATP, FAD, FMN, NAD, and STR; (B) AMP, GLC, HEM, and PO4. (C) Average ROC AUC values over all ligand types. Top-3 prediction success rate of (D) ATP, FAD, FMN, NAD, and STR; (E) AMP, GLC, HEM, and PO4. (F) Average Top-3 success rate over all ligand types. *NT in x-axis denotes experiments with no threshold used.

Mentions: Figure 4 shows the AUC values and the Top-3 success rate for different distance threshold values. To make individual curves more visible, the ligand types are arbitrarily divided into two groups that show similar trends: The first group contains pockets that bind to ATP, FAD, FMN, NAD, and STR (Figure 4A,D) while the another group includes pockets that bind to AMP, GLC, HEM, and PO4 (Figure 4B,E). In terms of the AUC value, ligand types in the first group (Figure 4A) tend to have higher values at the distance threshold between 0.15 and 0.25. On the other hand, the AUC values of the second group (Figure 4B) become higher as larger distance threshold values are used. This observation is consistent for the results with the Top-3 success rate (Figure 4D,E). Averaging the results of all the ligand types, the AUC values sharply increases until the threshold value of 0.2 and gradually increases as the threshold value is increased until the infinite distance was used (i.e., no threshold value used, NT) (Figure 4C). The average Top-3 success rate shows a similar trend, the value increases sharply until the threshold value of 0.2 and becomes stable after that point (Figure 4F). The largest Top-3 success rate is observed at the distance threshold of 0.30, which is 0.859.


Binding ligand prediction for proteins using partial matching of local surface patches.

Sael L, Kihara D - Int J Mol Sci (2010)

Prediction performance using shape and pocket size information. ROC AUC values of pockets that bind to (A) ATP, FAD, FMN, NAD, and STR; (B) AMP, GLC, HEM, and PO4. (C) Average ROC AUC values over all ligand types. Top-3 prediction success rate of (D) ATP, FAD, FMN, NAD, and STR; (E) AMP, GLC, HEM, and PO4. (F) Average Top-3 success rate over all ligand types. *NT in x-axis denotes experiments with no threshold used.
© Copyright Policy - open-access
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC3100846&req=5

f4-ijms-11-05009: Prediction performance using shape and pocket size information. ROC AUC values of pockets that bind to (A) ATP, FAD, FMN, NAD, and STR; (B) AMP, GLC, HEM, and PO4. (C) Average ROC AUC values over all ligand types. Top-3 prediction success rate of (D) ATP, FAD, FMN, NAD, and STR; (E) AMP, GLC, HEM, and PO4. (F) Average Top-3 success rate over all ligand types. *NT in x-axis denotes experiments with no threshold used.
Mentions: Figure 4 shows the AUC values and the Top-3 success rate for different distance threshold values. To make individual curves more visible, the ligand types are arbitrarily divided into two groups that show similar trends: The first group contains pockets that bind to ATP, FAD, FMN, NAD, and STR (Figure 4A,D) while the another group includes pockets that bind to AMP, GLC, HEM, and PO4 (Figure 4B,E). In terms of the AUC value, ligand types in the first group (Figure 4A) tend to have higher values at the distance threshold between 0.15 and 0.25. On the other hand, the AUC values of the second group (Figure 4B) become higher as larger distance threshold values are used. This observation is consistent for the results with the Top-3 success rate (Figure 4D,E). Averaging the results of all the ligand types, the AUC values sharply increases until the threshold value of 0.2 and gradually increases as the threshold value is increased until the infinite distance was used (i.e., no threshold value used, NT) (Figure 4C). The average Top-3 success rate shows a similar trend, the value increases sharply until the threshold value of 0.2 and becomes stable after that point (Figure 4F). The largest Top-3 success rate is observed at the distance threshold of 0.30, which is 0.859.

Bottom Line: The surface patches are encoded with the 3D Zernike descriptors.Unlike the existing methods which compare global characteristics of the protein fold or the global pocket shape, the local surface patch method can find functional similarity between non-homologous proteins and binding pockets for flexible ligand molecules.The proposed method improves prediction results over global pocket shape-based method which was previously developed by our group.

View Article: PubMed Central - PubMed

Affiliation: Department of Computer Science, Purdue University, West Lafayette, IN 47907, USA; E-Mail: lee399@cs.purdue.edu.

ABSTRACT
Functional elucidation of uncharacterized protein structures is an important task in bioinformatics. We report our new approach for structure-based function prediction which captures local surface features of ligand binding pockets. Function of proteins, specifically, binding ligands of proteins, can be predicted by finding similar local surface regions of known proteins. To enable partial comparison of binding sites in proteins, a weighted bipartite matching algorithm is used to match pairs of surface patches. The surface patches are encoded with the 3D Zernike descriptors. Unlike the existing methods which compare global characteristics of the protein fold or the global pocket shape, the local surface patch method can find functional similarity between non-homologous proteins and binding pockets for flexible ligand molecules. The proposed method improves prediction results over global pocket shape-based method which was previously developed by our group.

Show MeSH
Related in: MedlinePlus