This page lists papers submitted for IJCBDD that have been reviewed and
accepted but not yet published. Please note: article titles, authors,
abstracts and keywords may change upon publication.
Our TOC e-mail alerting service will notify you immediately when new issues
of IJCBDD are published on-line. Click here to register for our TOC E-Mail Alerting. We also offer the convenience of RSS
feeds which provide a means to view new content timely posted to your web
site or desktop. Click here to start to use our free RSS news feeds.
| 1. Improving Prediction Accuracy of Drug Activities by Utilizing Unlabelled Instances with Feature Selection |
| Authors: |
Guo-Zheng Li, Jack Y. Yang, Dan Li and Mary Qu Yang |
| Abstract: |
Molecular activities are critical for drug design, they can be predicted by QSAR (quantitative structure activity relationship) models, which overcomes the disadvantages of high cost and long cycle by employing the traditional experimental method. Because of the high cost of experiments, the number of drug molecules with known activity is much less than that of unknown, to predict molecular activities utilizing un-labelled instances will be an interesting issue. Here, semi-supervised learning (SSL) is introduced and a SSL method, Co-Training is investigated on predicting drug activities utilising unlabeled instances. At the same time, the numerous features extracted from the structures of drug molecules hurt prediction accuracy of QSAR models. Therefore, a novel algorithm called FESCOT is proposed, which applies feature selection to remove redundant and irrelevant features for Co-Training. Numerical experimental results on a data set of molecular
activities show that Co-Training improves prediction accuracy of molecular activities with unlabelled instances, and feature selection furthermore helps to improve the prediction ability of Co-Training. |
| Keywords: |
QSAR; semi-supervised learning; co-training; feature selection; K-nearest neighbour |
| |
| 2. Prediction of Lipid-Interacting Amino Acid Residues from Sequence Features |
| Authors: |
Liangjiang Wang, Stephanie J. Irausquin, Mary Qu Yang and Jack Y. Yang |
| Abstract: |
Proteins and lipids are integral components of cell membranes, and play important roles in cell signaling. Alternations of normal protein-lipid recognition may cause various diseases such as cancers and neurological disorders. However, molecular mechanisms underlying protein-lipid recognition are still poorly understood. In this study, we have developed a machine learning approach for prediction of lipid-interacting residues from amino acid sequence data. Protein sequences with known lipid-interacting residues were chosen from the Protein Data Bank (PDB). Support Vector Machines (SVMs) were then trained with data instances encoded with three biochemical features. The results suggest that lipid-interacting residues can be predicted from primary sequence information. To the best of our knowledge, this is the first study that applies machine learning to prediction of lipidinteracting residues based on amino acid sequence data. Our study provides useful information for understanding protein-lipid interactions, and may lead to advances in drug discovery. |
| Keywords: |
Lipid-Interacting Residues; Biochemical Features; Sequence-Based Prediction; Support Vector Machines; Machine Learning |
| |
| 3. A Comparison of algorithms for a complete backtranslation of oligopeptides |
| Authors: |
Mohieddine Missaoui, David R.C. Hill and Pierre Peyret |
| Abstract: |
When studying complex environments where the composing microorganisms are unknown, exploratory tools able to tackle with the biological diversity have to be proposed. DNA microarrays can be a good answer if we are in a position to propose all the probes that could target a specific enzyme. In addition, in the context of new metabolic pathways discovery, it appears that a full backtranslation of oligopeptides is also a promising approach. In both contexts it is preferable to have all the complete nucleic sequences corresponding to an enzyme of interest. In this paper, we revisit existing bioinformatics applications, which bring partial reverse translation solutions, and we compare two algorithms based on input oligopeptide degeneracy able to efficiently compute a complete backtranslation of oligopeptides. This kind of algorithms is precious for the discovery of new organisms and we show their performances on simulated and real biological datasets. |
| Keywords: |
Backtranslation; Microarrays; Oligopeptides; Probe Design |
| |
| 4. Detection and Prediction of Alternative Splicing in Arabidopsis thaliana |
| Authors: |
M. Park, D.L. Falcone and K.M. Daniels |
| Abstract: |
Alternative splicing is an important process for detecting the diversity arising from a single gene. Presently, most studies aimed at detecting alternatively spliced genes use ESTs (Expressed Sequence Tags). However, reliance on ESTs might have some weaknesses in predicting alternative splicing. First, the studies based on spliced transcripts analyze sequences by alignment rather than prediction and detection of sequence patterns. Second, EST libraries can be of uncertain quality. If a gene is expressed in a low level or is not expressed in a given tissue, then ESTs might not be available. To address these issues and to improve the quality of detection and prediction for alternative splicing, we propose a method that primarily uses pre-mRNAs. It is achieved by a decision tree algorithm using triplet nucleotides as attributes for each chromosome in Arabidopsis thaliana. Each decision tree shows that alternative and normal splicing have different splicing patterns according to triplet nucleotides. Based on the patterns, alternative splicing of unlabeled genes can also be predicted. In addition, we propose a novel algorithm for accurate prediction. It uses a formula based on triplet nucleotide consensus and decision tree levels. |
| Keywords: |
Alternative Splicing; pre-mRNA; Triplet Nucleotides; Detection
and Prediction; Arabidopsis thaliana |
| |
| 5. Inferring Transcription Factor Interactions Using a Novel HV-SVM Classifier |
| Authors: |
Xiao-Li Li, Jun-Xiang Lee, Bharadwaj Veeravalli and See-Kiong Ng |
| Abstract: |
Interactions between transcription factors (TFs) are necessary for deciphering the complex mechanisms of transcription regulation in eukaryotes. In this paper, we proposed a novel HV-kernel based SVM classifier (HV-SVM) to perform classification of TF-TF pairs based on their protein domain information and GO annotations. Specifically, two types of pairwise kernels, namely, a horizontal kernel and a vertical kernel, were combined to evaluate the similarity between a pair of TFs, and a Genetic algorithm was used to obtain kernel and feature weights to optimize the classifier’s performance. We applied our proposed HV-SVM method to predict TF interactions for Homo sapiens and Mus muculus. We obtained accuracy and F-measures of over 85% and an AUC of almost 93%. Such high quality of prediction demonstrates that HV-SVM is capable of making accurate predictions of TF-TF interactions even in the higher and more complex eukaryotes. |
| Keywords: |
transcription factor, Support Vector Machine; protein
domains; GO annotations
|
| |
| 6. Parsimony Accelerated Maximum Likelihood Searches |
| Authors: |
Kenneth Sundberg, Timothy O'Connor, Hyrum Carroll, Mark Clement and Quinn Snell |
| Abstract: |
Phylogenetic search is a key tool used in a variety of biological research endeavors. However, this search problem is known to be computationally difficult, due to the astronomically large search space, making the use of heuristic methods necessary. The performance of heuristic methods for finding maximum likelihood trees can be improved by using parsimony as an initial estimator for maximum likelihood. The time spent in performing the parsimony search to boost performance is insignificant compared to the time spent in the maximum likelihood search, leading to an overall gain in search time. These parsimony boosted maximum likelihood searches lead to topologies with scores statistically similar to the unboosted searches, but in less time. |
| Keywords: |
parsimony, maximum likelihood, phylogenetic search |
| |
| 7. Approximate Strip Exchanging |
| Authors: |
Swapnoneel Roy and Ashok Kumar Thakur |
| Abstract: |
Genome and other syntenic blocks rearrangements have become a topic of intensive study by phylogenists, comparative genomicists, and computational biologists: they are a feature of many cancers, must be taken into account to align highly divergent sequences, and constitute a phylogenetic marker of great interest. The mathematics of rearrangements is far more complex than for indels and mutations in sequences. Genome rearrangements have been modeled by a variety of primitives such as reversals, transpositions , block moves and block interchanges. We consider such a genome rearrangement primitive Strip Exchanges. Given a permutation, the challenge is to sort it by using minimum number of strip exchanges. A strip is a maximal substring of the permutation which is also a substring of the identity permutation. A strip exchanging move interchanges the positions of two chosen strips so that they merge with other strips. We present here the ¯rst non-trivial 2-approximation algorithm to this problem. For this purpose, we use the transposition graph model. We also observe that sorting by strip-exchanges, as well as sorting by block interchanges, are ¯xed-parameter-tractable. Lastly we discuss the application of strip exchanges in a di®erent area optical character recognition (OCR) with an example. |
| Keywords: |
Strip-Exchanges; Genome Rearrangement; Sorting Primitives; Approximation Algorithms. |
