Mutual Information Distance Research Articles

Identification of common microRNA-mRNA regulatory biomodules in human epithelial cancer

The complex regulatory network between microRNAs and gene expression remains unclear domain of active research. We proposed to address in part this complex regulation with a novel approach for the genome-wide identification of biomodules derived from paired microRNA and mRNA profiles, which could reveal correlations associated with a complex network of de-regulation in human cancer. Two published expression datasets for 68 samples with 11 distinct types of epithelial cancers and 21 samples of normal tissues were used, containing microRNA expression (Lu et al. Nature Letters 2005) and gene expression (Ramaswarmy et al. PNAS 2001) profiles, respectively. As results, the microRNA expression used jointly with mRNA expression can provide better classifiers of epithelial cancers against normal epithelial tissue than either dataset alone (p=1×10(-10), F-Test). We identified a combination of six microRNA-mRNA biomodules that optimally classified epithelial cancers from normal epithelial tissue (total accuracy = 93.3%; 95% confidence intervals: 86% - 97%), using penalized logistic regression (PLR) algorithm and three-fold cross-validation. Three of these biomodules are individually sufficient to cluster epithelial cancers from normal tissue using mutual information distance. The biomodules contain 10 distinct microRNAs and 98 distinct genes, including well known tumor markers such as miR-15a, miR-30e, IRAK1, TGFBR2, DUSP16, CDC25B and PDCD2. In addition, there is a significant enrichment (Fisher's exact test p=3×10(-10)) between putative microRNA-target gene pairs reported in five microRNA target databases and the inversely correlated micro-RNA-mRNA pairs in the biomodules. Further, microRNAs and genes in the biomodules were found in abstracts mentioning epithelial cancers (Fisher Exact Test, unadjusted p<0.05). Taken together, these results strongly suggest that the discovered microRNA-mRNA biomodules correspond to regulatory mechanisms common to human epithelial cancer samples. In conclusion, we developed and evaluated a novel comprehensive method to systematically identify, on a genome scale, microRNA-mRNA expression biomodules common to distinct cancers of the same tissue. These biomodules also comprise novel microRNA and genes as well as an imputed regulatory network, which may accelerate the work of cancer biologists as large regulatory maps of cancers can be drawn efficiently for hypothesis generation.

Image scrambling effect evaluation method based on mutual information distance of difference image

The new image scrambling effect evaluation method based on mutual information distance of difference image was proposed.The new concept of mutual information distance based on mutual information was firstly proposed,and then the difference image of the original image and scrambled image was obtained by means of difference operation on two images.At last the mutual information of two difference images was computed and acted as the new criteria of image scrambling effect evaluation.Experimental results show that the proposed method is effective to describe the relation between the scrambling effect and the number of iterations in the scrambling techniques,which is largely consistent with human vision.For different images,when some transformation is used,this evaluation method can reflect to some extent the scrambling effects in each scrambling stage.

Content-based similarity for 3D model retrieval and classification

With the rapid development of 3D digital shape information, content-based 3D model retrieval and classification has become an important research area. This paper presents a novel 3D model retrieval and classification algorithm. For feature representation, a method combining a distance histogram and moment invariants is proposed to improve the retrieval performance. The major advantage of using a distance histogram is its invariance to the transforms of scaling, translation and rotation. Based on the premise that two similar objects should have high mutual information, the querying of 3D data should convey a great deal of information on the shape of the two objects, and so we propose a mutual information distance measurement to perform the similarity comparison of 3D objects. The proposed algorithm is tested with a 3D model retrieval and classification prototype, and the experimental evaluation demonstrates satisfactory retrieval results and classification accuracy.

Gradient pattern analysis of short solar radio bursts

We analyze the weak component of the localized temporal pattern variability of 3 GHz solar burst observed by the Ondrejov radiospectrograph. A complex, short and weak impulsive sample from the time series was analyzed by applying a method based on the gradient pattern analysis and discrete wavelet decomposition. By analyzing canonical temporal variability patterns we show that the new method can reliably characterize the phenomenological dynamical process of short time series ( N ⩽ 10 3 measurements) as the radio burst addressed here. In the narrowest sense, by estimating the mutual information distance in the gradient spectra, we show that the fluctuation pattern of the short and weak 3 GHz impulsive solar burst, with energetic amplitudes <350 SFU, is closer to the intermittent and strong MHD turbulent variability pattern.

Evaluation of gene-expression clustering via mutual information distance measure.

BackgroundThe definition of a distance measure plays a key role in the evaluation of different clustering solutions of gene expression profiles. In this empirical study we compare different clustering solutions when using the Mutual Information (MI) measure versus the use of the well known Euclidean distance and Pearson correlation coefficient.ResultsRelying on several public gene expression datasets, we evaluate the homogeneity and separation scores of different clustering solutions. It was found that the use of the MI measure yields a more significant differentiation among erroneous clustering solutions. The proposed measure was also used to analyze the performance of several known clustering algorithms. A comparative study of these algorithms reveals that their "best solutions" are ranked almost oppositely when using different distance measures, despite the found correspondence between these measures when analysing the averaged scores of groups of solutions.ConclusionIn view of the results, further attention should be paid to the selection of a proper distance measure for analyzing the clustering of gene expression data.

On inequalities between mutual information and variation

We continue studying the relationship between mutual information and variational distance started in Pinsker's paper [1], where an upper bound for the mutual information via variational distance was obtained. We present a simple lower bound, which in some cases is optimal or asymptotically optimal. A uniform upper bound for the mutual information via variational distance is also derived for random variables with a finite number of values. For such random variables, the asymptotic behaviour of the maximum of mutual information is also investigated in the cases where the variational distance tends either to zero or to its maximum value.

SU‐DD‐A4‐01: Multi‐Day Multi‐Modality Image Co‐Registration

Purpose: To develop a methodology for multi‐day co‐registration of multi‐modality images taken throughout the course of radiotherapy, for the assessment of tumor response to treatment. Method and Materials: CT and PET image data sets of canines with various tumors were taken on a PET/CT scanner and were automatically co‐registered to each other by the scanner software. Using a commercial software package, ROI segmentation was performed on the CT data to define the tumor boundaries. Each PET data set was then masked by the ROI of its aligned CT image and then rigidly co‐registered with the masked PET data from other days. To align more specific uptake regions within the tumor, image warping was performed via landmarks placed on the co‐registered multi‐day PET data. Results: We found that ROIs can be roughly aligned by rigid registration of the outer tumor region as defined by the segmentation of CT data. Histogram correlation plots, aided by visual assessment, illustrated that normalized mutual information and Euclidean distance were the best‐suited criteria for rigid registration, and that increased accuracy in co‐registration was gained from image warping via landmarks placed on specific structures within the day‐to‐day rigidly co‐registered PET data. A review of warping techniques on the co‐registered PET data and generated data phantoms demonstrated that thin‐plate spline warping provided the most complete warping of landmarked structures, while flow warping was not as aggressive with non‐landmarked structures. Conclusion: Using this methodology of rigid registration of tumor boundaries defined from CT data, followed by inner feature warping of PET data proves to be an accurate and usable way to perform multi‐day multi‐modality image co‐registration. Further investigations include additional contouring of segmented regions, adding scaling parameters to the initial registrations, and refining landmark placement and warping techniques.

Entropy descriptors of the chemical bond in information theory. II. Applications to simple orbital models

The communication channels of information theory are generated for simple orbital models, in which each atom contributes a single atomic orbital (AO) to form the chemical bonds in the molecule. The 2- and 3-AO models represent the prototype chemical bonds in a diatomic molecule and in a symmetrical transition-state complex, respectively. The representative conjugated π bond systems (butadiene and benzene) in the Hückel approximation are also examined. Several entropy/information concepts introduced in Part I of this series (preceding article of this issue), including the conditional entropy, mutual information and information distance (entropy deficiency) quantities involving the input and output probability distributions of molecular communication systems in the atomic/orbital resolution, are used to characterize the chemical bond and its covalent and ionic components. The two-orbital model is used to generate representative atoms-in-molecules (AIM) information networks for both the orbital and spin probabilities of the ground- and excited-state configurations. The entropy ‘signatures’ of the model bonding, non-bonding, and antibonding electron configurations are discussed for both the orbital (atomic), spin, and orbital–spin probability distributions. A distinction between the electron-sharing and pair-sharing (coordination) bonds is explored and the entropic indices of the localized bonds are proposed. The overall bond-order conservation and a competition between the bond ionic and covalent components are examined in some detail. The entropy/information indices of the localized bond between a given pair of atoms in a polyatomic molecule are re-examined and tested on the model polyatomic systems.

Entropy descriptors of the chemical bond in information theory. I. Basic concepts and relations

The elements of the ‘communication theory’ of the chemical bond are outlined. The average uncertainties of the information theory characterizing the communication channels are summarized. The molecular system in atomic resolution is interpreted as the ‘communication’ channel, in which signals of the electron allocations to constituent atoms are propagated from the molecular input (‘source’), determined by the atomic ‘promolecule’, to the molecular output (‘receiver’), via the system chemicals bonds. This transmission network, connecting the free atoms of the promolecule reference and the bonded atoms-in-a-molecule, is determined by the molecular conditional two-electron probabilities in atomic resolution. Owing to the electron delocalization, the molecular information channels exhibit a chemical ‘noise’, which affects the transmission of the atom-assignment signals and the associated flow of information through the communication system. Several entropy/information concepts, including the conditional entropy, mutual information and information distance (cross-entropy, entropy deficiency), between the input and output probability distributions, are used to characterize the chemical bond and its covalent and ionic components. The conditional entropy and mutual information are identified as overall measures of the covalent and ionic bond components, respectively. Several reference input probabilities for probing different aspects of the chemical bond are examined. A distinction between the electron-sharing and the pair-sharing (coordination) bonds is explored and the entropic indices of the localized bonds are proposed. The average entropies of the local ‘stockholder’ communication channel, generated by the Hirshfeld partition of molecular electron densities, are proposed as complementary information descriptors of the chemical bond.