MUCKE aims to mine a large volume of images, to structure them conceptually and to use this conceptual structuring in order to improve large-scale image retrieval. The last decade witnessed important progress concerning low-level image representations. However, there are a number problems which need to be solved in order to unleash the full potential of image mining in applications. The central problem with low-level representations is the mismatch between them and the human interpretation of image content. This problem can be instantiated, for instance, by the incapability of existing descriptors to capture spatial relationships between the concepts represented or by their incapability to convey an explanation of why two images are similar in a content-based image retrieval framework. We start by assessing existing local descriptors for image classification and by proposing to use co-occurrence matrices to better capture spatial relationships in images. The main focus in MUCKE is on cleaning large scale Web image corpora and on proposing image representations which are closer to the human interpretation of images. Consequently, we introduce methods which tackle these two problems and compare results to state of the art methods. Note: some aspects of this deliverable are withheld at this time as they are pending review. Please contact the authors for a preview.

http://ieeexplore.ieee.org/iel2/4524/12820/00592460.pdf

Image Processing

Complex biological systems and cellular networks may underlie most genotype to phenotype relationships. Here, we review basic concepts in network biology, discussing different types of interactome networks and the insights that can come from analyzing them. We elaborate on why interactome networks are important to consider in biology, how they can be mapped and integrated with each other, what global properties are starting to emerge from interactome network models, and how these properties may relate to human disease.

Interactome networks and human disease

The objective of this study is to investigate and compare the results of three data mining approaches, the support vector machines (SVM), decision tree (DT), and Naive Bayes (NB) models for spatial prediction of landslide hazards in the Hoa Binh province (Vietnam). First, a landslide inventory map showing the locations of 118 landslides was constructed from various sources. The landslide inventory was then randomly partitioned into 70% for training the models and 30% for the model validation. Second, ten landslide conditioning factors were selected (i.e., slope angle, slope aspect, relief amplitude, lithology, soil type, land use, distance to roads, distance to rivers, distance to faults, and rainfall). Using these factors, landslide susceptibility indexes were calculated using SVM, DT, and NB models. Finally, landslide locations that were not used in the training phase were used to validate and compare the landslide susceptibility maps. The validation results show that the models derived using SVM have the highest prediction capability. The model derived using DT has the lowest prediction capability. Compared to the logistic regression model, the prediction capability of the SVM models is slightly better. The prediction capability of the DT and NB models is lower.

/pdf/landslide-susceptibility-assessment-in-vietnam-using-support-59qkb49nwd.pdf

Landslide Susceptibility Assessment in Vietnam Using Support Vector Machines, Decision Tree, and Naïve Bayes Models

Many transition-metal oxides show very large (“colossal”) magnitudes of the dielectric constant and thus have immense potential for applications in modern microelectronics and for the development of new capacitance-based energy-storage devices In the present work, we thoroughly discuss the mechanisms that can lead to colossal values of the dielectric constant, especially emphasising effects generated by external and internal interfaces, including electronic phase separation In addition, we provide a detailed overview and discussion of the dielectric properties of CaCu3Ti4O12 and related systems, which is today’s most investigated material with colossal dielectric constant Also a variety of further transition-metal oxides with large dielectric constants are treated in detail, among them the system La2−xSrxNiO4 where electronic phase separation may play a role in the generation of a colossal dielectric constant

/pdf/colossal-dielectric-constants-in-transition-metal-oxides-ipw2t6omi6.pdf

Colossal dielectric constants in transition-metal oxides

Predicting interactions between proteins and other biomolecules solely based on structure remains a challenge in biology. A high-level representation of protein structure, the molecular surface, displays patterns of chemical and geometric features that fingerprint a protein's modes of interactions with other biomolecules. We hypothesize that proteins participating in similar interactions may share common fingerprints, independent of their evolutionary history. Fingerprints may be difficult to grasp by visual analysis but could be learned from large-scale datasets. We present MaSIF (molecular surface interaction fingerprinting), a conceptual framework based on a geometric deep learning method to capture fingerprints that are important for specific biomolecular interactions. We showcase MaSIF with three prediction challenges: protein pocket-ligand prediction, protein-protein interaction site prediction and ultrafast scanning of protein surfaces for prediction of protein-protein complexes. We anticipate that our conceptual framework will lead to improvements in our understanding of protein function and design.

Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning.

Motivation: The limited availability of protein structures often restricts the functional annotation of proteins and the identification of their protein–protein interaction sites. Computational methods to identify interaction sites from protein sequences alone are, therefore, required for unraveling the functions of many proteins. This article describes a new method (PSIVER) to predict interaction sites, i.e. residues binding to other proteins, in protein sequences. Only sequence features (position-specific scoring matrix and predicted accessibility) are used for training a Naive Bayes classifier (NBC), and conditional probabilities of each sequence feature are estimated using a kernel density estimation method (KDE).

Results: The leave-one out cross-validation of PSIVER achieved a Matthews correlation coefficient (MCC) of 0.151, an F-measure of 35.3%, a precision of 30.6% and a recall of 41.6% on a non-redundant set of 186 protein sequences extracted from 105 heterodimers in the Protein Data Bank (consisting of 36 219 residues, of which 15.2% were known interface residues). Even though the dataset used for training was highly imbalanced, a randomization test demonstrated that the proposed method managed to avoid overfitting. PSIVER was also tested on 72 sequences not used in training (consisting of 18 140 residues, of which 10.6% were known interface residues), and achieved an MCC of 0.135, an F-measure of 31.5%, a precision of 25.0% and a recall of 46.5%, outperforming other publicly available servers tested on the same dataset. PSIVER enables experimental biologists to identify potential interface residues in unknown proteins from sequence information alone, and to mutate those residues selectively in order to unravel protein functions.

Availability: Freely available on the web at http://tardis.nibio.go.jp/PSIVER/

Contact:yoichi@nibio.go.jp; kenji@nibio.go.jp

Supplementary information:Supplementary data are available at Bioinformatics online.

/pdf/applying-the-naive-bayes-classifier-with-kernel-density-12lfw4qpqp.pdf

Applying the Naïve Bayes classifier with kernel density estimation to the prediction of protein–protein interaction sites

Summary: SHARP2 is a flexible web-based bioinformatics tool for predicting potential protein--protein interaction sites on protein structures. It implements a predictive algorithm that calculates multiple parameters for overlapping patches of residues on the surface of a protein. Six parameters are calculated: solvation potential, hydrophobicity, accessible surface area, residue interface propensity, planarity and protrusion (SHARP2). Parameter scores for each patch are combined, and the patch with the highest combined score is predicted as a potential interaction site. SHARP2 enables users to upload 3D protein structure files in PDB format, to obtain information on potential interaction sites as downloadable HTML tables and to view the location of the sites on the 3D structure using Jmol. The server allows for the input of multiple structures and multiple combinations of parameters. Therefore predictions can be made for complete datasets, as well as individual structures.

Availability: http://www.bioinformatics.sussex.ac.uk/SHARP2

Contact: s.jones@sussex.ac.uk

http://bioc.polytechnique.fr/biocomputing/papers/sharp2.pdf

SHARP2: protein--protein interaction predictions using patch analysis

Network analysis and in silico prediction of protein–protein interactions with applications in drug discovery

The PiRaNhA web server is a publicly available online resource that automatically predicts the location of RNA-binding residues (RBRs) in protein sequences. The goal of functional annotation of sequences in the field of RNA binding is to provide predictions of high accuracy that require only small numbers of targeted mutations for verification. The PiRaNhA server uses a support vector machine (SVM), with position-specific scoring matrices, residue interface propensity, predicted residue accessibility and residue hydrophobicity as features. The server allows the submission of up to 10 protein sequences, and the predictions for each sequence are provided on a web page and via email. The prediction results are provided in sequence format with predicted RBRs highlighted, in text format with the SVM threshold score indicated and as a graph which enables users to quickly identify those residues above any specific SVM threshold. The graph effectively enables the increase or decrease of the false positive rate. When tested on a non-redundant data set of 42 protein sequences not used in training, the PiRaNhA server achieved an accuracy of 85%, specificity of 90% and a Matthews correlation coefficient of 0.41 and outperformed other publicly available servers. The PiRaNhA prediction server is freely available at http://www.bioinformatics.sussex.ac.uk/PIRANHA.

/pdf/piranha-a-server-for-the-computational-prediction-of-rna-1pesux8f6s.pdf

PiRaNhA: a server for the computational prediction of RNA-binding residues in protein sequences

We have developed a method to predict ligand-binding sites in a new protein structure by searching for similar binding sites in the Protein Data Bank (PDB). The similarities are measured according to the shapes of the molecular surfaces and their electrostatic potentials. A new web server, eF-seek, provides an interface to our search method. It simply requires a coordinate file in the PDB format, and generates a prediction result as a virtual complex structure, with the putative ligands in a PDB format file as the output. In addition, the predicted interacting interface is displayed to facilitate the examination of the virtual complex structure on our own applet viewer with the web browser (URL: http://eF-site.hgc.jp/eF-seek).

/pdf/ef-seek-prediction-of-the-functional-sites-of-proteins-by-4hh5ml4qkp.pdf

Yoichi Murakami

Papers

Applying the Naïve Bayes classifier with kernel density estimation to the prediction of protein–protein interaction sites

SHARP2: protein--protein interaction predictions using patch analysis

Network analysis and in silico prediction of protein–protein interactions with applications in drug discovery

PiRaNhA: a server for the computational prediction of RNA-binding residues in protein sequences

eF-seek: prediction of the functional sites of proteins by searching for similar electrostatic potential and molecular surface shape.