Nathaniel Beagley

Evaluating the potential of multithreaded platforms for irregular scientific computations

The resurgence of current and upcoming multithreaded architectures and programming models led us to conduct a detailed study to understand the potential of these platforms to increase the performance of data-intensive, irregular scientific applications. Our study is based on a power system state estimation application and a novel anomaly detection application applied to network traffic data. We also conducted a detailed evaluation of the platforms using microbenchmarks in order to gain insight into their architectural capabilities and their interaction with programming models and application software. The evaluation was performed on the Cray MTA-2 and the Sun Niagar.

Combining Hierarchical and Associative Gene Ontology Relations With Textual Evidence in Estimating Gene and Gene Product Similarity

Two approaches have recently emerged where the similarity between two genes or gene products is obtained by comparing Gene Ontology (GO) annotations associated with the genes or gene products. One approach captures GO-based similarity in terms of hierarchical relations within each gene subontology, while the other relies on associative relations across the three gene subontologies. We propose a novel methodology where the two approaches can be merged and enriched by textual evidence extracted from biomedical literature with ensuing benefits in coverage and stronger correlation with sequence-based similarity

Cross-Ontological analytics: combining associative and hierarchical relations in the gene ontologies to assess gene product similarity

Gene and gene product similarity is a fundamental diagnostic measure in analyzing biological data and constructing predictive models for functional genomics. With the rising influence of the gene ontologies, two complementary approaches have emerged where the similarity between two genes/gene products is obtained by comparing gene ontology (GO) annotations associated with the gene/gene products. One approach captures GO-based similarity in terms of hierarchical relations within each gene ontology. The other approach identifies GO-based similarity in terms of associative relations across the three gene ontologies. We propose a novel methodology where the two approaches can be merged with ensuing benefits in coverage and accuracy.

Using the gene ontology to enrich biological pathways.

Most current approaches to automatic pathway generation are based on a reverse engineering approach in which pathway plausibility is solely derived from gene expression data and not independently validated. Alternative approaches use prior biological knowledge to validate automatically inferred pathways, but the prior knowledge is usually not sufficiently tuned to the pathology of focus. We present a novel pathway generation approach that combines insights from the reverse engineering and knowledge-based approaches to increase the biological plausibility of automatically generated regulatory networks and describe an application of this approach to transcriptional data from a mouse model of neuroprotection during stroke.

VIBE 2.0

Summary: Data fusion methods are powerful tools for evaluating experiments designed to discover measurable features of directly unobservable systems. We describe an interactive software platform, Visual Integration for Bayesian Evaluation, that ingests or creates Bayesian posterior probability matrices, performs data fusion and allows the user to interactively evaluate the classification power of fusing various combinations of data sources, such as transcriptomic, proteomics, metabolomics, biochemistry and function. Availability: http://omics.pnl.gov/software/VIBE.php Contact: bj@pnl.gov Supplementary information: Supplementary data are available at Bioinformatics online.

Probability Convergence in a Multithreaded Counting Application

The problem of counting specified combinations of a given set of variables arises in many statistical and data mining applications. To solve this problem, we introduce the PDtree data structure, which avoids exponential time and space complexity associated with prior work by allowing user specification of the tree structure. A straightforward parallelization approach using a Cray MTA-2 provides a speedup that is linear in the number of processors, but introduces nondeterminism into probability estimates. We prove a general convergence result that bounds the non-deterministic deviation of probability estimates relative to a sequential implementation. Beyond PDtrees, this convergence result applies to any counting application that takes a multithreaded streaming approach.

Enhancing Automatic Biological Pathway Generation with GO-Based Gene Similarity

Most current approaches to automatic pathway generation are based on a reverse engineering approach in which pathway plausibility is solely derived from microarray gene expression data. These approaches tend to lack in generality and offer no independent validation as they are too reliant on the pathway observables that guide pathway generation. By contrast, alternative approaches that use prior biological knowledge to validate pathways inferred from gene expression data may err in the opposite direction as the prior knowledge is usually not sufficiently tuned to the pathology of focus. In this paper, we present a novel pathway generation approach that combines insights from the reverse engineering and knowledge-based approaches to increase the biological plausibility of automatically generated regulatory networks.

Human-Centered Fusion Framework

In recent years the benefits of fusing signatures extracted from large amounts of distributed and/or heterogeneous data sources have been largely documented in various problems ranging from biological protein function prediction to cyberspace monitoring. In spite of significant progress in information fusion research, there is still no formal theoretical framework for defining various types of information fusion systems, defining and analyzing relations among such types, and designing information fusion systems using a formal method approach. Consequently, fusion systems are often poorly understood, are less than optimal, and/or do not suit user needs. To start addressing these issues, we outline a formal human-centered fusion framework for reasoning about fusion strategies. Our approach relies on a new taxonomy for fusion strategies, an alternative definition of information fusion in terms of parameterized paths in signature related spaces, an algorithmic formalization of fusion strategies and a library of numeric and dynamic visual tools measuring the impact as well as the impact behavior of fusion strategies. Using a real case of intelligence analysis we demonstrate that the proposed framework enables end users to rapidly 1) develop and implement alternative fusion strategies, 2) understand the impact of each strategy, 3) compare the various strategies, and 4) perform the above steps without having to know the mathematical foundations of the framework. We also demonstrate that the human impact on a fusion system is critical in the sense that small changes in strategies do not necessarily correspond to small changes in results.

Increasing the Efficiency of Data Storage and Analysis Using Indexed Compression

The massive data sets produced by the high- throughput, multidimensional mass spectrometry instruments used in proteomics create challenges in data acquisition, storage and analysis. Data compression can help mitigate some of these problems but at the cost of less efficient data access, which directly impacts the computational time of data analysis. We have developed a compression methodology that 1) is optimized for a targeted mass spectrometry proteomics data set and 2) provides the benefits of size and speed from compression while increasing analysis efficiency by allowing extraction of segments of uncompressed data from a file without having to uncompress the entire file. This paper describes our compression algorithm, presents comparative metrics of compression size and speed, and explores approaches for applying the algorithm to a generalized data set.

A Multi-Tier Provenance Model for Global Climate Research

Global climate researchers rely upon many forms of sensor data and analytical methods to help profile subtle changes in climate conditions. The U.S. Department of Energy Atmospheric Radiation Measurement (ARM) program provides researchers with curated Value Added Products (VAPs) resulting from continuous instrumentation streams, data fusion, and analytical profiling. The ARM operational staff and software development teams (data producers) rely upon a number of techniques to ensure strict quality control (QC) and quality assurance (QA) standards are maintained. Climate researchers (data consumers) are highly interested in obtaining as much provenance evidence as possible to establish data trustworthiness. Currently all the evidence is not easily attainable or identifiable without significant efforts to extract and piece together information from configuration files, log files, codes, or status information on the ARM website. Our objective is to identify a provenance model that serves the needs of both the VAP producers and consumers. This paper shares our initial results – a comprehensive multi-tier provenance model. We describe how both ARM operations staff and the climate research community can greatly benefit from this approach to more effectively assess and quantify the data historical record.