Fingerprint Kernels for Protein Structure Comparison

Fober, Thomas; Mernberger, Marco; Klebe, G.; Hüllermeier, Eyke

doi:10.1002/minf.201100149

Cited by 5 publications

(5 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A similar idea is used in CASSERT [33] and GPU-CASSERT [29], but structural residue descriptors consist of the information on relative position of the C α atom, the secondary structure type, and the residue type. Another feature-based approach is used in methods proposed by Fober and colleagues in [7,8] and Leinweber et al in CavSimBase [17]. These approaches make use of graphs to model molecular structures and represent various features of protein structures.…”

Section: Methods For Protein Structure Alignmentmentioning

confidence: 99%

High-throughput and scalable protein function identification with Hadoop and Map-only pattern of the MapReduce processing model

Mrozek

Suwała

2018

Knowl Inf Syst

View full text Add to dashboard Cite

Efficient computational solutions for identification of protein functions or finding structural homologs of proteins gain importance in the era of structural genomics and in the face of growing volumes of biological data. Structural alignments, which underlie these two processes, take a lot of time to complete, especially when performed for large collections of 3D protein structures. Fortunately, structural alignments can be carried out on well-separable and independent subsets of the whole macromolecular data repository, which perfectly fits the MapReduce processing paradigm of bringing computations to data. In this paper, we show how the protein function identification and finding structural homologs can be efficiently accelerated with the use of the MapReduce procedure executed on Hadoop cluster established in a virtualized compute environment or a private cloud. For this purpose, we propose Map-only processing pattern of the MapReduce procedure, which is formally defined in this paper. The solution that we show joins advantages of performing computations in small virtualized compute environments with large-scale computations in public clouds, thus allowing to perform structural alignments for a number of usage scenarios, including comparison of pairs of 3D protein structures during evaluation of predicted protein models, one-to-many comparisons while identifying possible functions of the given structure, or all-to-all alignments while investigating the divergence between known protein structures and classifying proteins by their fold. In this paper, we also present results of performance tests when scaling up nodes of the Hadoop cluster and increasing the degree of parallelism with the intention of improving efficiency of the computations. Keywords Bioinformatics • Big data • Proteins • Scalable computations and MapReduce • 3D protein structures • Cloud computing 148 D. Mrozek et al.

show abstract

Section: Methods For Protein Structure Alignmentmentioning

confidence: 99%

High-throughput and scalable protein function identification with Hadoop and Map-only pattern of the MapReduce processing model

Mrozek

Suwała

2018

Knowl Inf Syst

View full text Add to dashboard Cite

show abstract

“…Graph kernels work particularly well for small molecules such as ligands, but they are less useful for larger molecules such as proteins. They gave rather poor results in [36], which explains why we concentrated here on the maximum common subgraph as a representative for graph-based approaches.…”

Section: Similarity Measures For Cavitiesmentioning

confidence: 99%

“…As a third family of approaches, one can also represent the protein cavity as a feature vector, taking both the geometry of the cavity and physico-chemical properties into account -see e.g. [36,35]. Subsequently, traditional or specialized measures can be applied on these vectors to obtain similarity scores between protein cavities [56,57].…”

Section: Similarity Measures For Cavitiesmentioning

confidence: 99%

“…In addition to providing a complementary notion of protein families [26], these methods also allow for extracting relationships between cavities of unrelated proteins [27]. Similarity measures that highlight cavities and binding sites can be further subdivided into graph-based approaches such as [28,29,30,31,32], geometric approaches such as [33,34] and feature-based approaches such as [35,36]. These measures will be discussed more thoroughly in Section 2.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Identification of Functionally Related Enzymes by Learning-to-Rank Methods

Stock

Fober

Hüllermeier

et al. 2014

IEEE/ACM Trans. Comput. Biol. and Bioinf.

Self Cite

View full text Add to dashboard Cite

Enzyme sequences and structures are routinely used in the biological sciences as queries to search for functionally related enzymes in online databases. To this end, one usually departs from some notion of similarity, comparing two enzymes by looking for correspondences in their sequences, structures or surfaces. For a given query, the search operation results in a ranking of the enzymes in the database, from very similar to dissimilar enzymes, while information about the biological function of annotated database enzymes is ignored. In this work, we show that rankings of that kind can be substantially improved by applying kernel-based learning algorithms. This approach enables the detection of statistical dependencies between similarities of the active cleft and the biological function of annotated enzymes. This is in contrast to search-based approaches, which do not take annotated training data into account. Similarity measures based on the active cleft are known to outperform sequence-based or structure-based measures under certain conditions. We consider the Enzyme Commission (EC) classification hierarchy for obtaining annotated enzymes during the training phase. The results of a set of sizeable experiments indicate a consistent and significant improvement for a set of similarity measures that exploit information about small cavities in the surface of enzymes.

show abstract

“…Moreover, we would like to emphasize that the problem of comparing two protein binding sites is more difficult than the problem of comparing protein structures (e.g., at the fold level). Only relatively few works focus on the comparison of protein binding sites, which can be subdivided into feature-based [25], [26], [27], graphbased [8], [28], [29], [30], [31], [32] and geometric approaches [33], [34], [35], [36]. Geometric approaches are particularly powerful, since they do not rely on a lossy transformation from a set of points in Euclidean space to graphs or features, in contrast to the other approaches.…”

Section: Introductionmentioning

confidence: 99%

GPU-Based Point Cloud Superpositioning for Structural Comparisons of Protein Binding Sites

Leinweber

Fober

Freisleben

2018

IEEE/ACM Trans. Comput. Biol. and Bioinf.

Self Cite

View full text Add to dashboard Cite

In this paper, we present a novel approach to solve the labeled point cloud superpositioning problem for performing structural comparisons of protein binding sites. The solution is based on a parallel evolution strategy that operates on large populations and runs on GPU hardware. The proposed evolution strategy reduces the likelihood of getting stuck in a local optimum of the multimodal real-valued optimization problem represented by labeled point cloud superpositioning. The performance of the GPU-based parallel evolution strategy is compared to a previously proposed CPU-based sequential approach for labeled point cloud superpositioning, indicating that the GPU-based parallel evolution strategy leads to qualitatively better results and significantly shorter runtimes, with speed improvements of up to a factor of 1,500 for large populations. Binary classification tests based on the ATP, NADH, and FAD protein subsets of CavBase, a database containing putative binding sites, show average classification rate improvements from about 92 percent (CPU) to 96 percent (GPU). Further experiments indicate that the proposed GPU-based labeled point cloud superpositioning approach can be superior to traditional protein comparison approaches based on sequence alignments.

show abstract

Fingerprint Kernels for Protein Structure Comparison

Cited by 5 publications

References 37 publications

High-throughput and scalable protein function identification with Hadoop and Map-only pattern of the MapReduce processing model

High-throughput and scalable protein function identification with Hadoop and Map-only pattern of the MapReduce processing model

Identification of Functionally Related Enzymes by Learning-to-Rank Methods

GPU-Based Point Cloud Superpositioning for Structural Comparisons of Protein Binding Sites

Contact Info

Product

Resources

About