### Citations

1059 | Using bayesian network to analyze expression data
- Friedman, Linial, et al.
(Show Context)
Citation Context ... problem, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, =-=[10]-=-, [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for example [15], [16]), a compelling observatio... |

1017 | Mersenne Twister: A 623-dimensionally equidistributed uniform pseudorandom number generator
- Matsumoto, Nishimura
- 1998
(Show Context)
Citation Context ...congruential generators, typically used in the C standard library’s rand() and rand_r() functions, are fast but are not high quality sources of randomness [28] whereas the well-known Mersenne Twister =-=[29]-=- is higher quality, but slower, and needs to maintain a relatively large amount of state. The random shuffle kernel is run by multiple threads concurrently, thus the PRNG must maintain state for every... |

491 |
Elements of Information Theory. 2nd ed
- Cover, Thomas
- 2006
(Show Context)
Citation Context ...iables reflects the reduction in uncertainty of predicting the value of one random variable given the value of the other. Computing MI between random variables is a wellstudied topic in its own right =-=[23]-=-. TINGe follows the same general framework as ARACNe [6], but 2. Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tes... |

273 | A Shrinkage Approach to Large-Scale Covariance Matrix Estimation and Implications for Functional Genomics
- Schafer, Strimmer
- 2005
(Show Context)
Citation Context ...a Institute of Technology, Atlanta, USA. Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], =-=[4]-=-, [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for netwo... |

251 | Large-scale mapping and validation of escherichia coli transcriptional regulation from a compendium of expression profiles
- Faith, Hayete, et al.
(Show Context)
Citation Context ... USA. Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, [6], =-=[7]-=-, [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for example [... |

184 |
Estimating mutual information
- Kraskov, Stögbauer, et al.
- 2004
(Show Context)
Citation Context ...observed expression value of a gene X is replaced by its rank in the set of observed values of X . This transformation, called rank transformation, is considered a good approximation of homeomorphism =-=[24]-=-. Therefore, MI is computed on the rank transformed gene vectors instead, which are permutations over the integers 0, 1, 2, ...,m − 1. Hence, when operating on rank transformed vectors, a permutation ... |

170 |
Reverse engineering of regulatory networks in human B cells
- BASSO, MARGOLIN, et al.
- 2005
(Show Context)
Citation Context ...anta, USA. Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, =-=[6]-=-, [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for exam... |

165 |
Reverse engineering gene networks using singular value decomposition and robust regression
- Yeung, Tegner, et al.
- 2002
(Show Context)
Citation Context ...son correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. =-=[14]-=-, In works that assess different methods for network construction (see, for example [15], [16]), a compelling observation is that simpler methods such as those that only decipher pairwise linear inter... |

146 |
How to infer gene networks from expression profiles
- Bansal, Belcastro, et al.
(Show Context)
Citation Context ...], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for example =-=[15]-=-, [16]), a compelling observation is that simpler methods such as those that only decipher pairwise linear interactions are computationally expedient, whereas higher quality methods including mutual i... |

90 |
Revealing strengths and weaknesses of methods for gene network inference
- Marbach
- 2010
(Show Context)
Citation Context ..., [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for example [15], =-=[16]-=-), a compelling observation is that simpler methods such as those that only decipher pairwise linear interactions are computationally expedient, whereas higher quality methods including mutual informa... |

87 |
Random Numbers Fall Mainly in the Plane
- MARSAGLIA
- 1968
(Show Context)
Citation Context ... to perform the shuffle are unbiased. Linear congruential generators, typically used in the C standard library’s rand() and rand_r() functions, are fast but are not high quality sources of randomness =-=[28]-=- whereas the well-known Mersenne Twister [29] is higher quality, but slower, and needs to maintain a relatively large amount of state. The random shuffle kernel is run by multiple threads concurrently... |

85 |
Xorshift RNGs
- Panneton, Marsaglia, et al.
- 2003
(Show Context)
Citation Context ...multi-threaded rand_r() on a single coprocessor. Given the requirements of good quality randomness, small state, and high performance, we have chosen to use an Xorshift pseudo-random number generator =-=[30]-=-. We have implemented a vectorized version that uses the entire 512-bit width of the coprocessor’s vector register to generate 16 32-bit random numbers concurrently. The internal state of the PRNG is ... |

80 |
Estimating mutual information using B-spline functions–an improved similarity measure for analysing gene expression data
- Daub, Steuer, et al.
- 2004
(Show Context)
Citation Context ...gh sparsity of datasets is a problem for any method. Our work is based on TINGe [8], [9], a fast parallel network reconstruction technique that uses B-spline based mutual information (MI) computation =-=[20]-=-, data processing inequality for removing indirect interactions [6], and permutation testing for rigorous assessment of statistical significance [8]. This method is a distributed memory parallel algor... |

65 |
Sparse Graphical Gaussian Modeling of the Isoprenoid Gene Network in Arabidopsis Thaliana,” Genome Biology
- Wille, Zimmermann, et al.
- 2004
(Show Context)
Citation Context ...titute of Technology, Atlanta, USA. Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], =-=[5]-=-, information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network co... |

52 |
Estimation of mutual information using kernel density estimators
- YI, Rajagopalan, et al.
(Show Context)
Citation Context ...servations into a fixed number of bins b and then count the number of observations in each bin. This method is extremely fast but is imprecise and sensitive to the selection of boundaries of the bins =-=[25]-=-. To overcome this limitation, Daub et al. [20] have proposed to let each observation belong to k bins simultaneously with weights given by B-spline functions of order k. Let Bbk be a B-spline functio... |

48 |
Algorithm 235: Random permutation
- Durstenfeld
- 1964
(Show Context)
Citation Context ... time consuming kernel of TINGe is random shuffle that is used to permute observation vectors. The kernel takes an array and produces a random permutation of the array. We use the Durstenfeld version =-=[27]-=- of the FisherYates shuffle to perform the random permutation: for U from m-1 downto 1 do j = random int 0 <= j <= U exchange array[j] and array[i] This algorithm’s complexity is linear in the number ... |

30 |
An Arabidopsis gene network based on the graphical Gaussian model. Genome Res
- Ma, Gong, et al.
- 2007
(Show Context)
Citation Context ...eorgia Institute of Technology, Atlanta, USA. Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, =-=[3]-=-, [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for ... |

20 |
Using the principle of entropy maximization to infer genetic interaction networks from gene expression patterns
- Lezon
- 2006
(Show Context)
Citation Context ...iques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, =-=[13]-=-, and singular value decomposition. [14], In works that assess different methods for network construction (see, for example [15], [16]), a compelling observation is that simpler methods such as those ... |

20 | N.: Speeding up mutual information computation using nvidia cuda hardware
- SHAMS, BARNES
- 2007
(Show Context)
Citation Context ...blem of irregular data accesses, we expect MI to be faster than GMI in all cases. 8.1.2 Comparison with GPU based implementations There has been recent work on the parallelization of MI on GPUs [18], =-=[21]-=-, [22]. Out of these, CUDAMI [18] is the most recent and the fastest reported performance. Using Nvidia Tesla C2050 GPU, the authors report computing pairwise MI values for 10000 genes and 4000 observ... |

18 | Using Bayesian network inference algorithms to recover molecular genetic regulatory networks
- Yu, Smith, et al.
(Show Context)
Citation Context ...at simpler methods such as those that only decipher pairwise linear interactions are computationally expedient, whereas higher quality methods including mutual information [6] and Bayesian approaches =-=[17]-=- face scaling limitations. This has spurred research into the development of parallel methods [8], [9], [11], [12] and GPU implementations [18]. The focus of this paper is enabling the construction of... |

16 |
Arabidopsis gene co-expression network and its functional modules
- Mao
(Show Context)
Citation Context ...ience and Engineering, Georgia Institute of Technology, Atlanta, USA. Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, [1], =-=[2]-=-, Gaussian modeling, [3], [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that asses... |

14 |
Coexpression network based on natural variation in human gene expression reveals gene interactions and functions. Genome Res
- Nayak
- 2009
(Show Context)
Citation Context ...al Science and Engineering, Georgia Institute of Technology, Atlanta, USA. Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, =-=[1]-=-, [2], Gaussian modeling, [3], [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that ... |

9 | Mutual information computation and maximization using GPU
- Lin, Medioni
(Show Context)
Citation Context ...s work is useful in two additional ways. Mutual information is a widely used technique with its roots in information theory, and there has been recent work on its parallelization for GPUs [18], [21], =-=[22]-=-. Our work is the first implementation of a B-spline based mutual information kernel on the Intel Xeon Phi coprocessor, and is useful for porting other MI based applications to this architecture. Simi... |

8 |
Estimating genome-wide gene networks using nonparametric Bayesian network models on massively parallel computers
- Tamada, Imoto, et al.
(Show Context)
Citation Context ...em, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], =-=[11]-=-, [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for example [15], [16]), a compelling observation is t... |

6 | Reverse engineering and analysis of large genome-scale gene networks. Nucleic Acids Res 2013;41:e24
- Aluru, Zola, et al.
(Show Context)
Citation Context ...WX 0.107 0.149 0.865 0.993 7.304 With software prefetching Optimal 0.109 0.143 0.975 0.976 11.670 6 EXPERIMENTAL SETUP 6.1 Datasets used The dataset used in our experiments is taken from Aluru et al. =-=[9]-=-, which consists of expression data on the model plant Arabidopsis thaliana. A total of 3,546 non-redundant expression profiles measured using Affymetrix Arabidopsis ATH1 genechip were available from ... |

5 |
Parallel information-theory-based construction of genome-wide gene regulatory networks
- Zola, Aluru, et al.
- 2010
(Show Context)
Citation Context ... Given the importance of this problem, a number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, [6], [7], =-=[8]-=-, [9], Bayesian networks, [10], [11], [12], entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for example [15], ... |

5 |
Muller-Wittig W: Parallel mutual information estimation for inferring gene regulatory networks on GPUs
- Shi, Schmidt, et al.
(Show Context)
Citation Context ...including mutual information [6] and Bayesian approaches [17] face scaling limitations. This has spurred research into the development of parallel methods [8], [9], [11], [12] and GPU implementations =-=[18]-=-. The focus of this paper is enabling the construction of genome-scale gene regulatory networks on the Intel R© Xeon Phi TM coprocessor1 [19]. At this scale, the number of observations is often dwarfe... |

4 |
S.: Parallel bayesian network structure learning with application to gene networks
- Nikolova, Aluru
- 2012
(Show Context)
Citation Context ...number of mathematical techniques have been developed including Pearson correlation, [1], [2], Gaussian modeling, [3], [4], [5], information theory, [6], [7], [8], [9], Bayesian networks, [10], [11], =-=[12]-=-, entropy maximization, [13], and singular value decomposition. [14], In works that assess different methods for network construction (see, for example [15], [16]), a compelling observation is that si... |

3 |
A GPU-based implementation of differential evolution for solving the gene regulatory network model inference problem
- Ramirez-Chavez, Coello, et al.
(Show Context)
Citation Context ...TIONAL BIOLOGY AND BIOINFORMATICS 12 7.6 Comparison with GPU based implementations To the best of our knowledge, there are two implementations of gene regulatory network construction using GPUs [31], =-=[32]-=-. While Borelli et al. [31] use feature selection, Ramierz-Chavez et al. [32] use differential evolution for network inference. As mentioned in Section 1, several techniques have been used for the con... |

2 |
Accelerating gene regulatory networks inference through GPU/CUDA programming
- Borelli, Camargo, et al.
(Show Context)
Citation Context ...OMPUTATIONAL BIOLOGY AND BIOINFORMATICS 12 7.6 Comparison with GPU based implementations To the best of our knowledge, there are two implementations of gene regulatory network construction using GPUs =-=[31]-=-, [32]. While Borelli et al. [31] use feature selection, Ramierz-Chavez et al. [32] use differential evolution for network inference. As mentioned in Section 1, several techniques have been used for t... |