Parallel processing on large redundant biological data sets: Protein structures classification with CEPAR

Structural genomics provides an important approach for characterizing and understanding systems biology. As a step toward better integrating protein three-dimensional (3D) structural information in cancer systems biology, we have constructed a Human Cancer Pathway Protein Interaction Network (HCPIN) by analysis of several classical cancer-associated signaling pathways and their physical protein-protein interactions. Many well known cancer-associated proteins play central roles as "hubs" or "bottlenecks" in the HCPIN. At least half of HCPIN proteins are either directly associated with or interact with multiple signaling pathways. Although some 45% of residues in these proteins are in sequence segments that meet criteria sufficient for approximate homology modeling (Basic Local Alignment Search Tool (BLAST) E-value <10(-6)), only approximately 20% of residues in these proteins are structurally covered using high accuracy homology modeling criteria (i.e. BLAST E-value <10(-6) and at least 80% sequence identity) or by actual experimental structures. The HCPIN Website provides a comprehensive description of this biomedically important multipathway network together with experimental and homology models of HCPIN proteins useful for cancer biology research. To complement and enrich cancer systems biology, the Northeast Structural Genomics Consortium is targeting >1000 human proteins and protein domains from the HCPIN for sample production and 3D structure determination. The long range goal of this effort is to provide a comprehensive 3D structure-function database for human cancer-associated proteins and protein complexes in the context of their interaction networks. The network-based target selection (BioNet) approach described here is an example of a general strategy for targeting co-functioning proteins by structural genomics projects.

This study sought to identify a single molecule capable of managing all three manifestations of metabolic syndrome–hyperglycaemia, dyslipidaemia and hypertension. Two Protein Data Bank (PDB) depositions were selected and used to establish the baseline affinity that any designed molecule in this study should ideally exceed in order to be considered for further optimisation. These were PDB depositions 3VN2 and 2P54 describing the bound co-ordinates of the Peroxisome Proliferator Activated Receptor (PPAR)γ partial agonist and Angiotensin II Receptor (Ang(II)R) blocker telmisartan and of the experimental PPARα fibrate agonist GW590735 bound to their respective cognate receptors. These small molecules were extracted from their cognate receptors, docked into their non-cognate counterparts, conformational analysis performed, and the optimal conformers were selected as template scaffolds in two parallel processes. The first was a fragment based de novo approach. Here, molecular moieties from the optimal telmisartan and GW590735 scaffolds modelled in their non-cognate targets and considered critical to binding were identified and modelled, in order to produce seed structures capable of sustaining molecular growth at user-directed sites designated as H.spc atoms subsequent to their being docked within the non-cognate Ligand Binding Pockets (LBPs). The second approach was a Virtual Screening (VS) exercise. Here, the optimal telmisartan and GW590735 conformers were submitted as query molecules to VS databases both individually and in the form of a consensus pharmacophore. This VS exercise identified structurally diverse molecules which were electronically and spatially similar to the queries and which were capable of modulating the target receptors. The molecular cohorts identified through both VS and the de novo approaches were filtered for Lipinski Rule compliance. The molecules that survived filtering were then re-docked into the non-cognate PPARα and/or γ_LBPs, conformational analysis re-performed and the affinity of the optimal conformer measured for its cognate receptor quantified. Comparison was made to the baseline and non-cognate receptor affinities previously established, and the molecules exhibiting dual affinities exceeding baseline values were selected for further optimisation. The use of the “tried and tested” Ang(II)R blocker and fibrate scaffolds as templates predisposes to the identification of novel structures devoid of unacceptable toxicity.

As XML becomes increasingly popular, XML schema design has become an increasingly important issue. One of the central objectives of good schema design is to avoid data redundancies: redundantly stored information can lead not just only to a higher data storage cost but also to increased costs for data transfer and data manipulation. Furthermore, such data redundancies can lead to potential update anomalies, rendering the database inconsistent. One strategy to avoid data redundancies is to design redundancy-free schema from the start on the basis of known functional dependencies. We observe that XML databases are often "casually designed" and XML FDs may not be determined in advance. Under such circumstances, discovering XML data redundancies from the data itself becomes necessary and is an integral part of the schema refinement (or re-design) process. We present the design and implementation of the first system, DiscoverXFD, for efficient discovery of XML data redundancies. It employs a novel XML data structure and introduces a new class of partition-based algorithms. The XML data redundancies are defined on the basis of a new notion of XML functional dependency (XML FD) that (1) extends previous notions by incorporating set elements into the XML FD specification, and (2) maintains tuple-based semantics through the novel concept of Generalized Tree Tuple (GTT). Using this comprehensive XML FD notion, we introduce a new normal form (GTT-XNF) for XML documents, and provide comprehensive comparisons with previous studies. Given the set of data redundancies (in the form of redundancy-indicating XML FDs) discovered by DiscoverXFD, we describe a normalization algorithm for converting any original XML schema into one in GTT-XNF.

Advances in the areas of hardware technology, processor organization and programming science have given fresh impetus to research in parallel processing within single job streams. Given a program with a large degree of inherent parallelism, the theoretical time required for problem solution can be reduced “by a factor at least as great as the number of processors” [1].A need exists for new techniques for detecting parallel processable segments in sequentially coded programs and for indicating (representing) parallelism during actual program writing and coding. In a companion paper [4], the authors considered some methods of detecting and representing parallel processable “subtasks” within a task (process). In this paper we shall study the techniques of recognizing and representing parallel processable tasks within a single job stream (program).

This paper is concerned with the grain-size problem; how can we partition a program into tasks (concurrent modules) for parallel processing. Multiprocessor scheduling problem has long been studied by a number of researchers, and several scheduling algorithms have also been proposed. However, the effect of parallel processing strictly attributes of the given tasks such as the topology of task graph (directed acyclic graph representing precedence relations among tasks) and the balance of grain-size (computational time of each task) as well as the available number of processors. In order to improve the solution of multiprocessor scheduling problem, this paper proposes an excellent method for restructuring a given task graph. Analyzing the topology of the task graph, adequate tasks are divided one by one for increasing the parallelism of the task graph. The effect of task division is evaluated by scheduling tasks, where communication delays between distinct processors are considered. As a result, an appropriate task graph could be derived from the initially given task graph with respect to the designated number of processors. The proposed method is successfully applied to the parallel processing of robot-arm control computation on a multiprocessor system.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Mycobacterium tuberculosis, the causative agent of tuberculosis, has been a serious threat to public health. The recent emergence of extreme drug-resistant strains further worsened the situation and required more serious efforts to find a better cure for TB. Availability of the complete genome map of various Mycobacterium tuberculosis strains serves a basis to propose new therapeutic targets against Mycobacterium tuberculosis. The present study aimed to identify and characterize potential and novel drug target against Mycobacterium tuberculosis by using an in silico approach. The study consisted of three phases. In phase I, subtractive/comparative genomics was performed on genomic data of 17 strains of Mycobacterium tuberculosis comprised of 66,988 protein sequences. The analysis resulted in three identical hypothetical proteins of distinct origin. The shortlisted sequences were non-homologous to the host, a prerequisite characteristic of potential drug target. One of the sequences, i.e., WP_003901628.1 (the locus tag name is CFBR_RS17985) from the CDC5180 strain was further subjected to other bioinformatics methods in phase II. In phase II, the shortlisted protein sequence was qualitatively characterized for subcellular localization, functional family prediction and evolutionary relationships. The predicted subcellular localization of the query protein was found in cytoplasmic region. The functional family classification resulted in classifying the protein in transferase superfamily. Finally, in phase III of the study, the structure-based methods (i.e., homology modeling and docking) were performed to build the three-dimensional (3D) structure and to predict the putative binding site. The PDB-ID 3DR4 was found to be a suitable template having 30.1% sequence identity. The binding site of modeled protein was deciphered with DoGSite scorer tool. Later on, the ProBis tool was used to probe structurally similar binding sites in Protein Data Bank. Subsequently, the co-crystal ligands of the Protein Data Bank structures (prioritized by high-rank Z scores) were extracted and cross-docked into the binding site of the modeled protein. The cross-docking results revealed two ligands, i.e., 4′-deoxy-4′-aminopyridoxal-5′-phosphate and OG1 from two Protein Data Bank structures, i.e., 1MDO and 4K2M as best docked with RMSDs 2.8 and 2.5 A, respectively. It is inferred from the protein structure that the identified protein is alanine-rich with a PI < 7. Current study convincingly proposed that the shortlisted hypothetical protein (WP_003901628.1) is a potential drug target against which new drug candidates could be designed in a hope to cure tuberculosis.

Molecular biology experiments are generating an unprecedented amount of information from a variety of different experimental modalities. DNA sequencing machines, proteomics mass cytometry and microscopes generate huge amounts of data every day. Not only is the data large, but it is also multidimensional. Understanding trends and getting actionable insights from these data requires techniques that allow comprehension at a high level but also insight into what underlies these trends. Lots of small errors or poor summarization can lead to false results and reproducibility issues in large data sets. Hence it is essential we do not cherry-pick results to suit a hypothesis but instead examine all data and publish accurate insights in a data-driven way. This article will give an overview of some of the problems faced by the researcher in understanding epigenetic changes (which are related to changes in the physical structure of DNA) when presented with raw analysis results using visualization methods. We will also discuss the new challenges faced by using machine learning which can be helped by visualization.

We present novel optimization-based classification models that are general purpose and suitable for developing predictive rules for large heterogeneous biological and medical data sets. Our predictive model simultaneously incorporates (1) the ability to classify any number of distinct groups; (2) the ability to incorporate heterogeneous types of attributes as input; (3) a high-dimensional data transformation that eliminates noise and errors in biological data; (4) the ability to incorporate constraints to limit the rate of misclassification, and a reserved-judgment region that provides a safeguard against over-training (which tends to lead to high misclassification rates from the resulting predictive rule); and (5) successive multi-stage classification capability to handle data points placed in the reserved judgment region. Application of the predictive model to a broad class of biological and medical problems is described. Applications include: the differential diagnosis of the type of erythemato-squamous diseases; genomic analysis and prediction of aberrant CpG island meythlation in human cancer; discriminant analysis of motility and morphology data in human lung carcinoma; prediction of ultrasonic cell disruption for drug delivery; identification of tumor shape and volume in treatment of sarcoma; multistage discriminant analysis of biomarkers for prediction of early atherosclerosis; fingerprinting of native and angiogenic microvascular networks for early diagnosis of diabetes, aging, macular degeneracy and tumor metastasis, and prediction of protein localization sites. In all these applications, the predictive model yields correct classification rates ranging from 80% to 100%. This provides motivation for pursuing its use as a medical diagnostic, monitoring and decision-making tool.

InterMine is a data integration warehouse and analysis software system developed for large and complex biological data sets. Designed for integrative analysis, it can be accessed through a user-friendly web interface. For bioinformaticians, extensive web services as well as programming interfaces for most common scripting languages support access to all features. The web interface includes a useful identifier look-up system, and both simple and sophisticated search options. Interactive results tables enable exploration, and data can be filtered, summarized, and browsed. A set of graphical analysis tools provide a rich environment for data exploration including statistical enrichment of sets of genes or other entities. InterMine databases have been developed for the major model organisms, budding yeast, nematode worm, fruit fly, zebrafish, mouse, and rat together with a newly developed human database. Here, we describe how this has facilitated interoperation and development of cross-organism analysis tools and reports. InterMine as a data exploration and analysis tool is also described. All the InterMine-based systems described in this article are resources freely available to the scientific community.

Summary form only given. The general process of automatic program parallelization, first consists of determining the source program dependencies, then determining a parallel execution order of all the program operations. This turns out to determine a scheduling and/or an allocation, i.e. a function which associates to each operation, an execution date, and/or a processor number. Such a function is known as temporal transformation if it is a scheduling, a spatial one if it is an allocation, and a space/time transformation, if it is a combination of both. Then, synchronizations must be inserted to satisfy the dependencies between the statements assigned to different processors. Finally, the last stage of parallelization is the generation of the target parallel program, respecting the chosen transformation, and taking into account the target machine characteristics (processor number, memory architecture, etc.). There are many scheduling and allocation methods for polyhedral programs. However, few efficient methods have been devoted to code generation, namely those generating code from allocation. As the latter has generally the advantage of speed and good locality, we are interested in designing an allocation based method of code generation for a loop nest. The generated program has the structure of a nest involving an outer parallel loop enumerating processors and enclosing, in a serial order, the set of statements executed by the same processor.

Complete and highly redundant data sets were collected at nine different wavelengths between 0.80 and 2.65 A on a xenon derivative of porcine pancreatic elastase in both air and helium atmospheres. The magnitude of the anomalous signal, as assessed by the xenon-peak height in the anomalous difference Patterson synthesis, is affected by the wavelength of data collection as well as by the scaling model used. For data collected at wavelengths longer than 1.7 A, the use of a three-dimensional scaling protocol is essential in order to obtain the highest possible anomalous signal. Based on the scaling protocols currently available, the optimal wavelength range for data collection appears to be between 2.1 and 2.4 A. Beyond that, any further increase in signal will be compensated for or even superseded by a concomitant increase in noise, which cannot be fully corrected for. Data collection in a helium atmosphere yields higher I/sigma(I) values, but not significantly better anomalous differences, than data collection in air.

ALPHORN: A TOOLBOX FOR FAULT-TOLERANT, DISTRIBUTED PROCESS CONTROL SYSTEMS

The application of wireless robotic inspection vehicles equipped with different NDE payloads has been introduced previously, with emphasis placed on inspection applications in hazardous and inaccessible environments. A particular challenge to the practical application of such robotic inspection lies in the localization of the devices. The authors here consider a probabilistic approach to both the positioning and defect problems by using the location of the robot and the NDE measurements (acquired from the onboard transducers) to make inference about defect existence and position. Using a particle filter approach running locally on the robots, the vehicle location is tracked by fusing noisy redundant data sets supplying positional information.

Image matching in large scale environments is challenging due to the large number of features used in typical representations. In this paper we investigate methods for reducing the number of SIFT (Scale invariant feature transform) features in an image based localization application. We find that reductions of up to 59% in the number of features can result in improved performance of a naive matching algorithm for highly redundant data sets. However, those improvements do not carry over to visual bag of words, where a more moderate feature reduction (up to 16%) is often needed to maintain performance similar to the non-reduced set. Our reduced features have performed better than other robust feature descriptors namely HoG, GIST and ORB on all data sets with naive matching. The main contribution of this paper is the compact feature representation of a large scale environment for robust 2D image matching.

Fortran code for SU(3) lattice gauge theory with and without MPI checkerboard parallelization