Abstract

The problem of search and retrieval of images using relevance feedback has attracted tremendous attention in recent years from the research community. A real-world-deployable interactive image retrieval system must (1) be accurate, (2) require minimal user-interaction, (3) be efficient, (4) be scalable to large collections (millions) of images, and (5) support multi-user sessions. For good accuracy, we need effective methods for learning the relevance of image features based on user feedback, both within a user-session and across sessions. Efficiency and scalability require a good index structure for retrieving results. The index structure must allow for the relevance of image features to continually change with fresh queries and user-feedback. The state-of-the-art methods available today each address only a subset of these issues. In this paper, we build a complete system FISH - Fast Image Search in Huge databases. In FISH, we integrate selected techniques available in the literature, while adding a few of our own. We perform extensive experiments on real datasets to demonstrate the accuracy, efficiency and scalability of FISH. Our results show that the system can easily scale to millions of images while maintaining interactive response time.

Abstract

An important problem in biological data analysis is to pre-dict the family of a newly discovered sequence like a pro-tein or DNA sequence, using the collection of available se-quences. In this paper we tackle this problem and presentREBMEC, a Repeat Based Maximum Entropy Classifier of biological sequences. Maximum entropy models are known to be theoretically robust and yield high accuracy,but are slow. This makes them useful as benchmarks to evaluate other classifiers. Specifically, REBMEC is based on the classical Generalized Iterative Scaling (GIS) al-gorithm and incorporates repeated occurrences of subse-quences within each sequence. REBMEC uses maximal frequent subsequences as features but can support other types of features as well. Our extensive experiments on two collections of protein families show that REBMEC performs as well as existing state-of-the-art probabilistic classifiers for biological sequences without using domain-specific background knowledge such as multiple align-ment, data transformation and complex feature extraction methods. The design of REBMEC is based on genericideas that can apply to other domains where data is orga-nized as collections of sequences.

Abstract

In this paper, we present RBNBC, a Repeat Based Naive Bayes Classifier of bio-sequences that uses maximal fre-quent subsequences as features. RBNBC’s design is based on generic ideas that can apply to other domains where the data is organized as collections of sequences. Specifically,RBNBC uses a novel formulation of Naive Bayes that incor-porates repeated occurrences of subsequences within each sequence. Our extensive experiments on two collections of protein families show that it performs as well as existing state-of-the-art probabilistic classifiers for bio-sequences.This is surprising as it is a pure data mining based generic classifier that does not require domain-specific background knowledge. We note that domain-specific ideas could fur-ther increase its performance.

Abstract

Hoogsteen-sugar base pattern of RNA base pairing is studied using density functional theory. Hydrogen bonding patterns of these base apirs are characterized using NBO analysis and AIM analysis. Correlation between strenghth of base pairing and nature of donor-acceptor combinations is also carried out.

Abstract

Predicting the 3D native conformation of a protein given the amino acid sequence is known as protein structure prediction (PSP)problem and is one of the greatest challenge of computational biology. The current work uses a real valued Genetic Algorithm (GA), a powerful variant of conventional GA to simulate the PSP problem. It consists of three evolutionary operators to manipulate the genes and a fitness function based on a simple force field. The individuals of the population correspond to different conformations of the same polypeptide chain, represented by a string of (phi,psi)torsion angles. The conformations are generated under the constraints of Ramachandran plot along with secondary structure information, which are then screened through a set of knowledge based biophysical filters, viz. persistence length and radius of gyration. Only the structures satisfying the filtering criteria are considered for energy minimization using GA. The algorithm has been validated on a set of known globular protein containing 2-4 secondary structure elements. For all the test proteins the algorithm converges rapidly ad the converged structure shows a backbone RMSD (root mean square deviation) or 3-6A as compared to the native structure.

Abstract

Motivation: Various types of functional RNAs-tRNA, Ribosomes, Introns, Aptamers, Ribozymes Diverse and Dynamic molecules Recent availability of RNA crystal structures has facilitated study and analysis

Abstract

Schistosoma mansoni, a trematode parasite causes schistosomiasis, a tropical disease affecting more than 200 million people. It lives in an aerobic environment and therefore has effective redox mechanisms for its survival as well as to evade reactive oxygen species from its host. Although, the host performs these activities by two different systems: thioredoxin and glutaredoxin separately, the parasite has only one unique multifunctional enzyme, thioredoxin glutathione reductase (TGR) for both the activities which is its main bottleneck. Therefore this emerges as a promising target for schistosomiasis. TGR is a homodimer of 65kDa sub units. each consisting 598 amino acids with a selenocysteine residue in the C-terminus. It shows an unprecedented fusion of two proteins: glutaredoxin (Grx) and thioredoxin reductase (TR). Although many biochemical studies have been reported for TGR, the three dimensional structural information is still missing. Herein, we present some of our efforts towards development of a homology model for S. mansoni TGR (SmTGR) using Modeller9.1. A BLAST search with SmTGR sequence revealed about 60% homology with different mammalian TRs and the initial 100 residues showed close similarity with Grx. Therefore, homology modeling was carried out using both TR and Grx templates. The developed models were dimerized and validated by Procheck analysis. The validated structure was then minimized by MD simulations using heated simulated annealing method.

Abstract

Purine riboswitches constitute a class of highly structured regions within the 5'-untranaslated sequence stretches of mRNAs that specifically bind purines to regulate the gene expression. A unique feature of the ligand binding in these riboswitches is the larger role of hydrogen bonding compared to stacking [1]. In order to study the inherent strength and specificity of molecular recognition involved in these interactions, we have carried out gas phase quantum chemical studies on the metabolite binding pockets of adenine [2] and guanine [2] binding riboswitches, with adenine, guanine and a modified pyrimidine respectively, at the B3LYP/6-31G level of theory. An analysis of optimized structures reveals that for three conserved residues - C74, U47 and U51, the ligand-base hydrogen bonding pattern remains consereved. On the other hand, the U22 residue deviates from its crystal geometary, suggesting that its is held in the crystal geometry through strong backbone and local environmental constraints. We have carried out Morokuma decomposition and interaction energy evaluations as well as NBO and AIM analysis to understand the strength and nature of binding of the individual aptamer residues with their respective purine metabolites. The dominant features of these non-covalent interactions have also been analyazed. Acknowledgements: AM thanks DBT for research grant. PS and SS thank CSIR, New Delhi for JRFs. [1] Gilbert, S. D.; Stoddard, C. D.; Wise, S. J.; Batey, R. T. J. Mol. Biol. 2006, 359, 754-768 [2] Serganov, A.; Yuan, Y.; Pikovskaya, O.; Polonskaia, A.; Malinina, L.; Phan, A. T.; Hobartner, C.; Micura, R.; Breakar, R. R.; Patel, D. J. Chem. Biol. 2004, 11, 1729-1741.

Abstract

The need for computational characterization of DNA structure in the context of DNA bending/binding, B ->A transitions in protein-DNA complexes and also in the context of strand orientation especially in DNA triplexes and quadruplexes have given rise to several fine grained parameters, such as backbone conformations, sugar pucker and chi angle, local strand orientation, base pair and base pair step parameters backbone base inclination etc. With their unpaired bases, extensive non helical regions, non-canonical base pairs and tertiary interactions, RNA molecules host a large variety of recurrent structural and functional motifs and parameters such as Zp, defined in the context of double helical DNA base pairs and base pair steps and effectively used for A and B DNA differentiations, can not be used for computationally driven automated mining of complex RNA motifs. Though several algorithms have been proposed based on backbone conformational parameters and though algorithms for computational detection of base pairs and analysis of base pair parameters, are available, there are no reports of motif mining studies involving parameters which, as with Zp, simultaneously involve backbone conformation as well as base pairing geometry. In this work, we propose a novel set of parameters, Omega Pseudo-Torsions and Omega Distances, which capture backbone geometries at base(i) - base(j) interaction points and which can be used to computationally characterize structural features both in RNA and DNA. Pseudo torsion angles: Omega eta: P(i)-C4'(j)-P(j); Omega theta: P(i+1)-C4'(j)-P(j+1) Omega 1: P(i)-C4'(i)-C4'(j)-P(j+1); Omega 2: P(i+1)-C4'(i)-C4'(j)-P(j) Pseudo bond distance: Omega distance: C4'(i)-C4'(j) Where 'i' and 'j' are bases interacting through hydrogen bonds. We report the results of our benchmarking studies of Omega parameters with standard DNA structure classification parameters and of our preliminary studies on their effectiveness in classifying local geometries of RNA strands in different base pairing (canonical and non-canonical) contexts.

Abstract

DNA structures and their variations are important in several contexts such as chromatin remodeling and transcriptional initiation and regulation etc. Theneed for computational characterization of DNA structures associated with DNA bending/binding, B ? A transitions in protein-DNA complexes, molecular dynamics trajectory analysis and also in the context of strand orientation, especially in DNA triplexes and quadruplexes, have given rise to several fine grained parameters, such as backbone conformations, sugar pucker and ? angle, local strand orientation, base pair and base pair step parameters, backbone base inclination etc. With their unpaired bases, extensive non helical regions, non-canonical base pairs and tertiary interactions, RNA molecules host a large variety of recurrent structural and functional motifs and parameters such as zp, defined in the context of double helical DNA base pairs and base pair steps and effectively used for A and B DNA differentiations, can not be used for computationally driven automated mining of complex RNA motifs. Thus for RNA structures, though several motif mining algorithms based on backbone conformational parameters and involving computational detection and analysis of base pairs and their parameters are available, there are no reports of studies involving parameters which simultaneously involve backbone conformation as well as base pairing geometry. In this work, we propose a novel set of parameters, Omega Torsions and Omega Distances, which captures backbone geometries at base(i) ? base(j) interaction points and which can be used to computationally characterize structural features both in RNA and DNA. Pseudo torsion angles: Omega ?: P(i)-C4'(i)-C4'(j)-P(j); Omega ?: P(i+1)-C4'(i)-C4'(j)-P(j+1) Omega 1: P(i)-C4'(i)-C4'(j)-P(j+1); Omega 2: P(i+1)-C4'(i)-C4'(j)-P(j) Pseudo bond distance: Omega distance: C4'(i)-C4'(j) Where 'i' and 'j' are bases interacting through hydrogen bonds. We report the results of our benchmarking studies of Omega parameters, as applied to protein DNA complexes including e.g. TATA box-TBP complexes and nucleosomal DNA, with standard DNA structure classification parameters, such as zp, and of our preliminary studies on their effectiveness in classifying local geometries of RNA strands in different base pairing contexts. We conclude that omega parameters can potentially lead to the development of a single set of parameters which can be used to characterize both DNA as well as RNA structures.