S. Durga Bhavani

Analysis of E.coli promoter recognition problem in dinucleotide feature space

MOTIVATION Patterns in the promoter sequences within a species are known to be conserved but there exist many exceptions to this rule which makes the promoter recognition a complex problem. Although many complex feature extraction schemes coupled with several classifiers have been proposed for promoter recognition in the current literature, the problem is still open. RESULTS A dinucleotide global feature extraction method is proposed for the recognition of sigma-70 promoters in Escherichia coli in this article. The positive data set consists of sigma-70 promoters with known transcription starting points which are part of regulonDB and promec databases. Four different kinds of negative data sets are considered, two of them biological sets (Gordon et al., 2003) and the other two synthetic data sets. Our results reveal that a single-layer perceptron using dinucleotide features is able to achieve an accuracy of 80% against a background of biological non-promoters and 96% for random data sets. A scheme for locating the promoter regions in a given genome sequence is proposed. A deeper analysis of the data set shows that there is a bifurcation of the data set into two distinct classes, a majority class and a minority class. Our results point out that majority class constituting the majority promoter and the majority non-promoter signal is linearly separable. Also the minority class is linearly separable. We further show that the feature extraction and classification methods proposed in the paper are generic enough to be applied to the more complex problem of eucaryotic promoter recognition. We present Drosophila promoter recognition as a case study. AVAILABILITY http://202.41.85.117/htmfiles/faculty/tsr/tsr.html.

SMOTE Based Protein Fold Prediction Classification

Protein contact maps are two dimensional representations of protein structures. It is well known that specific patterns occuring within contact maps correspond to configurations of protein secondary structures. This paper addresses the problem of protein fold prediction which is a multi-class problem having unbalanced classes. A simple and computationally inexpensive algortihm called Eight-Neighbour algortihm is proposed to extract novel features from the contact map. It is found that of Support Vector Machine (SVM) which can be effectively extended from a binary to a multi-class classifier does not perform well on this problem. Hence in order to boost the performance, boosting algorithm called SMOTE is applied to rebalance the data set and then a decision tree classifier is used to classify “folds” from the features of contact map. The classification is performed across the four major protein structural classes as well as among the different folds within the classes. The results obtained are promising validating the simple methodology of boosting to obtain improved performance on the fold classification problem using features derived from the contact map alone.

Application of neural networks for protein sequence classification

Protein sequence classification is modelled as a binary classification problem where an unlabeled protein sequence is checked to see if it belongs to a known set of protein superfamilies or not. In this paper we used multilayer perceptrons with supervised learning algorithm to learn the binary classification. The training data consists of two sets-a positive set belonging to an identified set of protein superfamily and a negative set comprising sequences from other superfamilies. When applying neural networks the first problem to be addressed is feature extraction. In this paper we used the new feature extraction techniques proposed by Wang et al. Simulations reveal that the neural network is able to classify with good precision for myosin and photochrome superfamilies in the data set that we have chosen as positive. Also the results for globin superfamily are good, thus validating the methodology of feature extraction and the application of neural networks for protein sequence classification as suggested by Wang et al. But, for Actin and Ribonuclease superfamilies the network showed poor performance. One possible reason for this may be that the choice of sequences in the negative data set is not optimal. We conclude from this work that the classification performance depends upon a proper selection of sequences for positive and negative data sets.

Symmetric encryption using logistic map

In Symmetric cryptography both the sender and receiver agree with the same key before they start communicating secretly. Same key is used for encryption and decryption. The communication is secret as long as the key is kept secret. A desirable property of symmetric encryption is termed as avalanche effect by which two different keys produces different ciphertext for the same message. Essential properties of chaos functions are sensitivity to initial conditions and ergodicity, which makes two nearby keys to generate different cipher texts. Sensitivity to initial conditions property of chaos can be exploited to produce avalanche effect. We propose a symmetric encryption algorithm which uses logistic map. We show that the keys with negligible difference generate different cipher texts. Cryptanalysis of the proposed algorithm shows that it is resistant to various attacks and stronger than existing encryption algorithms.

Graph Isomorphism Detection Using Vertex Similarity Measure

Measures of vertex similarity have been incorporated in graph matching algorithms. Graph matching tries to retrieve a 1-1 correspondence between vertices of two given graphs. In this paper, the vertex similarity measure of Blondel et al. is studied for its usefulness in detecting graph isomorphism. Firstly, the applicability of this measure to distinguish similar pairs from dissimilar pairs is shown to be limited in scope even for small graphs. In a preliminary experiment, we show that Blondel’s vertex similarity measure does not retrieve the isomorphism within a graph of 14 nodes. We propose a refinement of Blondel’s measure. Zager et al. also refine Blondel’s measure and further propose a graph matching algorithm. We propose a graph matching algorithm based on the lines of Zager et al. and test our algorithm against Zager’s as well as Blondel’s and show that the proposed refinement performs better than both the measures with regard to graph isomorphism problem. The performance is evaluated systematically on a large bench mark data set made available by Foggia et al. The proposed algorithm performs with 90.10% accuracy on all of the 18,200 pairs of isomorphic graphs available in the benchmark dataset.

Temporal probabilistic measure for link prediction in collaborative networks

Link prediction addresses the problem of finding potential links that may form in the future. Existing state of art techniques exploit network topology for computing probability of future link formation. We are interested in using Graphical models for link prediction. Graphical models use higher order topological information underlying a graph for computing Co-occurrence probability of the nodes pertaining to missing links. Time information associated with the links plays a major role in future link formation. There have been a few measures like Time-score, Link-score and T_Flow, which utilize temporal information for link prediction. In this work, Time-score is innovatively incorporated into the graphical model framework, yielding a novel measure called Temporal Co-occurrence Probability (TCOP) for link prediction. The new measure is evaluated on four standard benchmark data sets : DBLP, Condmat, HiePh-collab and HiePh-cite network. In the case of DBLP network, TCOP improves AUROC by 12 % over neighborhood based measures and 5 % over existing temporal measures. Further, when combined in a supervised framework, TCOP gives 93 % accuracy. In the case of three other networks, TCOP achieves a significant improvement of 5 % on an average over existing temporal measures and an average of 9 % improvement over neighborhood based measures. We suggest an extension to link prediction problem called Long-term link prediction, and carry out a preliminary investigation. We find TCOP proves to be effective for long-term link prediction.Link prediction addresses the problem of finding potential links that may form in the future. Existing state of art techniques exploit network topology for computing probability of future link formation. We are interested in using Graphical models for link prediction. Graphical models use higher order topological information underlying a graph for computing Co-occurrence probability of the nodes pertaining to missing links. Time information associated with the links plays a major role in future link formation. There have been a few measures like Time-score, Link-score and T_Flow, which utilize temporal information for link prediction. In this work, Time-score is innovatively incorporated into the graphical model framework, yielding a novel measure called Temporal Co-occurrence Probability (TCOP) for link prediction. The new measure is evaluated on four standard benchmark data sets : DBLP, Condmat, HiePh-collab and HiePh-cite network. In the case of DBLP network, TCOP improves AUROC by 12 % over neighborhood based measures and 5 % over existing temporal measures. Further, when combined in a supervised framework, TCOP gives 93 % accuracy. In the case of three other networks, TCOP achieves a significant improvement of 5 % on an average over existing temporal measures and an average of 9 % improvement over neighborhood based measures. We suggest an extension to link prediction problem called Long-term link prediction, and carry out a preliminary investigation. We find TCOP proves to be effective for long-term link prediction.

Heterogeneous link prediction based on multi relational community information

Social networks consisting of edges annotated with multiple links are natural models for real-world networks and pose a challenge for network analysis. Link prediction, predicting future links or missing links in a multi-relational network, is an important task from applications perspective. In large networks, time and memory are major constraints for link prediction. In this context, an algorithm is proposed to improve upon the recent solutions proposed for this problem. In this paper, a parallel method for predicting links in heterogeneous networks is proposed. As social networks exhibit a natural community structure and the nodes interact more within community than with the nodes in other communities, this multi relational community information is used for parallelization. Utilizing the existing state-of-the art algorithms for multi-relational link prediction as well as community discovery algorithms, the proposed method, computes multi relational link prediction scores in each community. The results of implementation of these algorithms on bench-mark data sets show that community information does significantly help in improving the performance of multi-relational link prediction.

Design of Secure Chaotic Hash Function Based on Logistic and Tent Maps

The main contribution of the paper is two-fold: Building step by step a chaotic hash function starting from a weak but basic algorithm and analyzing its strengths and weaknesses in order to make it stronger. We start with a basic chaotic hash function with a 128-bit message digest based on Baptista’s encryption algorithm. In the next steps, a pseudo-random number generator using chaotic tent map is incorporated within the hash algorithm and perturbation and block chaining approaches are used to strengthen the hash function. In the literature on chaotic cryptography we have not seen preimage and second-preimage resistance analyis being done which we carry out and show that the proposed hash function is strong against both these attacks. Further, the standard collision analysis is performed in the space of 1-bit neighbourhood of a given message. The hash function is shown to exhibit diffusion effect with average hamming distance between message digests obtained as 63 bits which is close to the ideal of 50%. The collision performance compares favourably with that of chaotic hash functions proposed in the recent literature. It is to be emphasized that the existing chaotic hash functions in the literature use a multitude of chaotic maps whereas we show in this paper that using two chaotic maps judiciously achieves a secure hash function.

Enhancement to community-based multi-relational link prediction using co-occurrence probability feature

Predicting future links or missing links is one of the useful application tasks in the analysis of social networks. Time and memory are major challenges for the link prediction task in large multi-relational social networks. This challenge is addressed in this paper, by proposing a parallel method for link prediction. Community information is used for parallelization since social networks tend to form natural communities, and probability of intra community node interaction is much more than inter community interaction. For prediction task, along with the standard topological features, the recently proposed local probabilistic graph model is also used. This model infers the joint co-occurrence probability of two nodes (i, j) from Markov Random Field constructed using the nodes in the neighbourhood of (i, j). In this paper, we adopt the supervised framework of MR-HPLP of the literature by including the co-occurrence probability feature in the multi relational environment, and reducing the dimensionality of the feature vector. This method is evaluated on a challenging benchmark multi relational dataset and COP is shown to outperform as an unsupervised predictor. Further MR-HPLP-COP shows significant improvement in AUROC as well as AUPR scores over all the ten existing predictors on the benchmark data set. In particular, MR-HPLP-COP shows AUROC of over 90% for two data sets for which the existing methods give a prediction performance of around 75%.

Study of Diversity and Similarity of Large Chemical Databases Using Tanimoto Measure

ZINC is a freely available chemical database which contains 27 million compounds including Drug-like, Natural Products, FDA etc., along with 9 molecular features. In this paper firstly we compute an additional number of 49 molecular features and represent the entire chemical space in the 58-length finger print space. Tanimoto metric, a popular similarity measure is used to mine the chemical space for extracting similar and diverse fingerprints. One of the important issues is that of choosing a proper reference string. Experiments with different reference strings are carried out to assess the appropriateness of a reference string. A finger print which is constituted by mandating non-trivial presence of each feature is found to be the best. Further a method which is independent of reference string is proposed using pairwise distribution but this raises the time complexity from linear to quadratic. A subgoal of this paper is also to propose a scheme that extracts a small sample data set that reflects the similarity and diversity of the population. Towards this, we conduct stratified sampling of Natural Products Database(NPD) which has 90,000 chemical compounds by dividing the space along strata representing distinct structures (rings) and then compute pairwise similarity profile. This scheme can be extended to other data bases that reside in ZINC.