Areas Of Computational Biology Research Articles

MFmap: A semi-supervised generative model matching cell lines to tumours and cancer subtypes.

Translating in vitro results from experiments with cancer cell lines to clinical applications requires the selection of appropriate cell line models. Here we present MFmap (model fidelity map), a machine learning model to simultaneously predict the cancer subtype of a cell line and its similarity to an individual tumour sample. The MFmap is a semi-supervised generative model, which compresses high dimensional gene expression, copy number variation and mutation data into cancer subtype informed low dimensional latent representations. The accuracy (test set F1 score >90%) of the MFmap subtype prediction is validated in ten different cancer datasets. We use breast cancer and glioblastoma cohorts as examples to show how subtype specific drug sensitivity can be translated to individual tumour samples. The low dimensional latent representations extracted by MFmap explain known and novel subtype specific features and enable the analysis of cell-state transformations between different subtypes. From a methodological perspective, we report that MFmap is a semi-supervised method which simultaneously achieves good generative and predictive performance and thus opens opportunities in other areas of computational biology.

Challenges and limitations in the studies of glycoproteins: A computational chemist's perspective.

Experimenters face challenges and limitations while analyzing glycoproteins due to their high flexibility, stereochemistry, anisotropic effects, and hydration phenomena. Computational studies complement experiments and have been used in characterization of the structural properties of glycoproteins. However, recent investigations revealed that computational studies face significant challenges as well. Here, we introduce and discuss some of these challenges and weaknesses in the investigations of glycoproteins. We also present requirements of future developments in computational biochemistry and computational biology areas that could be necessary for providing more accurate structural property analyses of glycoproteins using computational tools. Further theoretical strategies that need to be and can be developed are discussed herein.

Constructing Unrooted Phylogenetic Trees with Reinforcement Learning

With the development of sequencing technologies, more and more amounts of sequence data are available. This poses additional challenges, such as processing them is usually a complex and time-consuming computational task. During the construction of phylogenetic trees, the relationship between the sequences is examined, and an attempt is made to represent the evolutionary relationship. There are several algorithms for this problem, but with the development of computer science, the question arises as to whether new technologies can be exploited in these areas of computational biology.
 In the following publication, we investigate whether the reinforced learning model of machine learning can generate accurate phylogenetic trees based on the distance matrix.

Chemical graph generators.

Chemical graph generators are software packages to generate computer representations of chemical structures adhering to certain boundary conditions. Their development is a research topic of cheminformatics. Chemical graph generators are used in areas such as virtual library generation in drug design, in molecular design with specified properties, called inverse QSAR/QSPR, as well as in organic synthesis design, retrosynthesis or in systems for computer-assisted structure elucidation (CASE). CASE systems again have regained interest for the structure elucidation of unknowns in computational metabolomics, a current area of computational biology.

Protein Fold Prediction for Protein Sequences of Low Identity Based on Evolutionary and Spatial Features Using Random Forest Algorithm

Protein fold prediction is a milestone step towards predicting protein tertiary structure from protein sequence. It is considered one of the most researched topics in the area of Computational Biology. It has applications in the area of structural biology and medicines. Extracting sensitive features for prediction is a key step in protein fold prediction. The actionable features are extracted from keywords of sequence header and secondary structure representations of protein sequence. The keywords holding species information are used as features after verifying with uniref100 dataset using TaxId. Prominent patterns are identified experimentally based on the nature of protein structural class and protein fold. Global and native features are extracted capturing the nature of patterns experimentally. It is found that keywords based features have positive correlation with protein folds. Keywords indicating species are important for observing functional differences which help in guiding the prediction process. SCOPe 2.07 and EDD datasets are used. EDD is a benchmark dataset and SCOPe 2.07 is the latest and largest dataset holding astral protein sequences. The training set of SCOPe 2.07 is trained using 93 dimensional features vector using Random forest algorithm. The prediction results of SCOPe 2.07 test set reports the accuracy of better than 95%. The accuracy achieved on benchmark dataset EDD is better than 93%, which is best reported as per our knowledge.

An Application of Random Walk Resampling to Phylogenetic HMM Inference and Learning.

Statistical resampling methods are widely used for confidence interval placement and as a data perturbation technique for statistical inference and learning. An important assumption of popular resampling methods such as the standard bootstrap is that input observations are identically and independently distributed (i.i.d.). However, within the area of computational biology and bioinformatics, many different factors can contribute to intra-sequence dependence, such as recombination and other evolutionary processes governing sequence evolution. The SEquential RESampling ("SERES") framework was previously proposed to relax the simplifying assumption of i.i.d. input observations. SERES resampling takes the form of random walks on an input of either aligned or unaligned biomolecular sequences. This study introduces the first application of SERES random walks on aligned sequence inputs and is also the first to demonstrate the utility of SERES as a data perturbation technique to yield improved statistical estimates. We focus on the classical problem of recombination-aware local genealogical inference. We show in a simulation study that coupling SERES resampling and re-estimation with recHMM, a hidden Markov model-based method, produces local genealogical inferences with consistent and often large improvements in terms of topological accuracy. We further evaluate method performance using empirical HIV genome sequence datasets.

Functional module extraction from gene expression data using data mining techniques

A set of correlated and co-expressed genes, often referred as a functional module, play a synergistic role during any disease or any biological activities. Genes participating in a common module may cause clinically similar diseases and shares the common genetic origin of their associated disease phenotypes. Identifying such modules may be helpful in system level understanding of biological and cellular processes or pathophysiologic basis of associated diseases. As a result, detecting such functional modules is an active research issue in the area of computational biology. Many techniques have been proposed so far to find functional modules based on gene co-regulation or co-expression data. These methods are broadly categorized into nonnetwork based gene expression clustering techniques and network-based methods that extract modules from gene co-expression networks using expression data sources. We surved main approaches for obtaining modules, and we evaluated their performance regarding finding biologically significant gene modules in the light of both microarray and RNASeq data. No prior effort, other than independent assessment, has been made so far to evaluate their performances in an integrated way in the light of both microarray and RNASeq data. It could be observed that these methods are basically based on certain features and several other features are ignored. No single technique appears to be effective in all respect. Therefore, there is a possibility that some significant modules might be missed out. Keeping this in view we came up with a solution which would engulf the goodness of all the methods into one. We proposed a multilayer ensemble approach based on few well-known module detection techniques into one. We observed that ensemble of techniques enhances the quality of modules in terms biological significance. We evaluated the effectiveness of the ensemble approach in detecting disease specific modules. Often, a set of genes found to be responsible for dual (or even more) functionalities while participating in multiple overlapping module formation. A more compact form of overlapping module structure is intrinsic structure, where a set of genes within a module playing additional role despite its parent role where it belongs to. We proposed a unique way of detecting such module structures by using the ensemble of modules obtained by the subspace clustering. We used the concept of frequent itemset mining, a step in Association Rule Mining, to derive such compact modules as subspace clusters. To the best of our knowledge, no prior work is attempted so far to detect overlapping and intrinsic modules simultaneously from ensemble outcomes. Finally, we proposed a ranking method to infer disease responsible key genes based on gene expression data. We used top ranked disease significant modules derived by our ensemble method and based on significance of the module with respect to disease pathways. We applied our ranking method in Breast Cancer and Alzheimer's Disease (AD). We inferred top genes and assessed their significance related to the disease with the help of various gene-disease association databases. Experimental results revealed that BRCA 1 , BRCA 2 , PTEN, ABI 1 and CASP8 are the top key genes in Breast Cancer, whereas, MAPK 1 , APP, CASP 7 , APOE and PSEN 1 are the key players in AD.

Remote homology detection using GA and NSGA-II on physicochemical properties

Remote homology detection at amino acid level is a complex problem in the area of computational biology. We have used machine learning algorithms to predict the homology of un-annotated protein sequences which can save time and cost. This work is divided in three phases. Initially the features are extracted from protein sequences using Principal Component Analysis (PCA) to build a chromosome set with representative features of each protein based on physicochemical properties. Second stage involves GA for the construction of a set of chromosomes for classification based on PCA and initialises the classifier to build up an error matrix. Third stage uses NSGA-II, crossover and mutation, and tournament selection for the next set of chromosomes. The output of this experiment is a set of minimum classification error values and minimum number of features used for classification of protein families. This approach gives superior accuracy over the profile-based methods.

Remote homology detection using GA and NSGA-II on physicochemical properties

Remote homology detection at amino acid level is a complex problem in the area of computational biology. We have used machine learning algorithms to predict the homology of un-annotated protein sequences which can save time and cost. This work is divided in three phases. Initially the features are extracted from protein sequences using Principal Component Analysis (PCA) to build a chromosome set with representative features of each protein based on physicochemical properties. Second stage involves GA for the construction of a set of chromosomes for classification based on PCA and initialises the classifier to build up an error matrix. Third stage uses NSGA-II, crossover and mutation, and tournament selection for the next set of chromosomes. The output of this experiment is a set of minimum classification error values and minimum number of features used for classification of protein families. This approach gives superior accuracy over the profile-based methods.

Qualitative assessment of functional module detectors on microarray and RNASeq data

A set of correlated and co-expressed genes, often referred as a functional module, play a synergistic role during any disease or any biological activities. Genes participating in a common module may cause clinically similar diseases and share a common genetic origin of their associated disease phenotypes. Identifying such modules may be helpful in system-level understanding of biological and cellular processes or pathophysiologic basis of associated diseases. As a result detecting such functional modules is an active research issue in the area of computational biology. Some techniques have been proposed so far to find functional modules based on gene co-regulation or co-expression data. These methods are broadly categorized into non-network based gene expression clustering techniques and network-based methods that extract modules from gene co-expression networks using expression data sources. We survey main approaches for obtaining modules, and we evaluate their performance regarding finding biologically significant gene modules in light of both microarray and RNASeq data. No prior effort, other than independent assessment, has been made so far to evaluate their performances in an integrated way in the light of both microarray and RNASeq data. We assess the significance of the modules in terms of gene ontology and pathway analysis. We select a few of the best performers to access their capability in finding disease-specific modules. Our comparison reveals that no single algorithm is a winner in all respects. Moreover, performances vary widely with microarray and RNASeq data. Relatively, biclustering performs better, when we consider microarray expression data, but fails to perform well in case of RNASeq data. Network-based techniques work better in RNASeq.

CORE: A Software Tool for Delineating Regions of Recurrent DNA Copy Number Alteration in Cancer.

Collections of genomic intervals are a common data type across many areas of computational biology. In cancer genomics, in particular, the intervals often represent regions with altered DNA copy number, and their collections exhibit recurrent features, characteristic of a given cancer type. Cores of Recurrent Events (CORE) is a versatile computational tool for identification of such recurrent features. Here we provide practical guidance for the use of CORE, implemented as an eponymous R package.

Intelligent computational model for classification of sub-Golgi protein using oversampling and fisher feature selection methods.

Golgi is one of the core proteins of a cell, constitutes in both plants and animals, which is involved in protein synthesis. Golgi is responsible for receiving and processing the macromolecules and trafficking of newly processed protein to its intended destination. Dysfunction in Golgi protein is expected to cause many neurodegenerative and inherited diseases that may be cured well if they are detected effectively and timely. Golgi protein is categorized into two parts cis-Golgi and trans-Golgi. The identification of Golgi protein via direct method is very hard due to limited available recognized structures. Therefore, the researchers divert their attention toward the sequences from structures. However, owing to technological advancement, exploration of huge amount of sequences was reported in the databases. So recognition of large amount of unprocessed data using conventional methods is very difficult. Therefore, the concept of intelligence was incorporated with computational model. Intelligence based computational model obtained reasonable results, but the gap of improvement is still under consideration. In this regard, an intelligent automatic recognition model is developed in order to enhance the true classification rate of sub-Golgi proteins. In this approach, discrete and evolutionary feature extraction methods are applied on the benchmark Golgi protein datasets to excerpt salient, propound and variant numerical descriptors. After that, an oversampling technique Syntactic Minority over Sampling Technique is employed to balance the data. Hybrid spaces are also generated with combination of these feature spaces. Further, Fisher feature selection method is utilized to reduce the extra noisy and redundant features from feature vector. Finally, k-nearest neighbor algorithm is used as learning hypothesis. Three distinct cross validation tests are used to examine the stability and efficiency of the proposed model. The predicted outcomes of proposed model are better than the existing models in the literature so far. Finally, it is anticipated that the proposed model will provide the foundation to pharmaceutical industry in drug design and research community to innovate new ideas in the area of computational biology and bioinformatics.

A Hybrid Multiobjective Memetic Metaheuristic for Multiple Sequence Alignment

Over the last 25 years, the multiple sequence alignment (MSA) problem has attracted the attention of biologists because it is one of the major techniques used in several areas of computational biology, such as homology searches, genomic annotation, protein structure prediction, gene regulation networks, or functional genomics. This problem implicates the alignment of more than two biological sequences, and is considered as a nondeterministic polynomial time optimization problem. In this paper, we find a number of different approaches for dealing with this biological sequence alignment problem. Basically, we distinguish six main groups: 1) exact methods; 2) progressive methods; 3) consistency-based methods; 4) iterative methods; 5) evolutionary algorithms; and 6) structure-based methods. In this paper, we propose the use of evolutionary computation and multiobjective optimization for solving this bioinformatics problem. A multiobjective version of a memetic metaheuristic is presented: hybrid multiobjective metaheuristics for MSA. In order to prove the effectiveness of the new proposal, we use three structure-based benchmarks created by using empirical data as input. The results obtained by our method are compared with well-known methods published in this paper, concluding that the new approach presents remarkable accuracy when dealing with sets of sequences with a low sequence similarity, the most frequent ones in real world.

SuMoTED: An intuitive edit distance between rooted unordered uniquely-labelled trees

Defining and computing distances between tree structures is a classical area of study in theoretical computer science, with practical applications in the areas of computational biology, information retrieval, text analysis, and many others. In this paper, we focus on rooted, unordered, uniquely-labelled trees such as taxonomies and other hierarchies. For trees as these, we introduce the intuitive concept of a ‘local move’ operation as an atomic edit of a tree. We then introduce SuMoTED, a new edit distance measure between such trees, defined as the minimal number of local moves required to convert one tree into another. We show how SuMoTED can be computed using a scalable algorithm with quadratic time complexity. Finally, we demonstrate its use on a collection of music genre taxonomies.

Using a Human Drug Network for generating novel hypotheses about drugs

Analyzing different drugs for various purposes is an important issue in the area of computational biology. We cate- gorize the previous computational studies into Individual and Network approaches. While the Individual approach focuses on one specific drug without considering its relationship with other drugs, the Network approach considers also the drugs relation- ships. In this paper, we apply a Network approach, previously proposed for discovering the relationships among diseases, to drug data. We construct a Human Drug Network (HDN) for 200 different drugs based on functional and structural information available in the PPI network. For evaluating our proposed HDN, first, we analyzed the literature to prove that the proposed HDN is biologically meaningful. Second, we used the HDN to augment the initial prior knowledge of different drugs. As an example of prior knowledge, we considered the initial seed proteins (a set of proteins which are previously known to be drug targets) of each drug. We clustered the HDN nodes using the Markov CLustering Algorithm (MCL) and then, we augmented the seed proteins of each drug based on the cluster it belongs to. In the end, we concluded that our proposed HDN enables us to generate novel hypotheses (in terms of potential drug target proteins) and produce complementary results comparing to existing methods.

Identification of microRNA precursor with the degenerate K-tuple or Kmer strategy

The microRNA (miRNA), a small non-coding RNA molecule, plays an important role in transcriptional and post-transcriptional regulation of gene expression. Its abnormal expression, however, has been observed in many cancers and other disease states, implying that the miRNA molecules are also deeply involved in these diseases, particularly in carcinogenesis. Therefore, it is important for both basic research and miRNA-based therapy to discriminate the real pre-miRNAs from the false ones (such as hairpin sequences with similar stem-loops). Most existing methods in this regard were based on the strategy in which RNA samples were formulated by a vector formed by their Kmer components. But the length of Kmers must be very short; otherwise, the vector's dimension would be extremely large, leading to the “high-dimension disaster” or overfitting problem. Inspired by the concept of “degenerate energy levels” in quantum mechanics, we introduced the “degenerate Kmer” (deKmer) to represent RNA samples. By doing so, not only we can accommodate long-range coupling effects but also we can avoid the high-dimension problem. Rigorous jackknife tests and cross-species experiments indicated that our approach is very promising. It has not escaped our notice that the deKmer approach can also be applied to many other areas of computational biology. A user-friendly web-server for the new predictor has been established at http://bioinformatics.hitsz.edu.cn/miRNA-deKmer/, by which users can easily get their desired results.

A COMPARATIVE STUDY OF PROTEIN TERTIARY STRUCTURE PREDICTION METHODS

Protein structure prediction (PSP) from amino acid sequence is one of the high focus problems in bioinformatics today. This is due to the fact that the biological function of the protein is determined by its three dimensional structure. The understanding of protein structures is vital to determine the function of a protein and its interaction with DNA, RNA and enzyme. Thus, protein structure is a fundamental area of computational biology. Its importance is intensed by large amounts of sequence data coming from PDB (Protein Data Bank) and the fact that experimentally methods such as X-ray crystallography or Nuclear Magnetic Resonance (NMR)which are used to determining protein structures remains very expensive and time consuming. In this paper, different types of protein structures and methods for its prediction are described.

Model Checking CSL for Markov Population Models

Markov population models (MPMs) are a widely used modelling formalism in the area of computational biology and related areas. The semantics of a MPM is an infinite-state continuous-time Markov chain. In this paper, we use the established continuous stochastic logic (CSL) to express properties of Markov population models. This allows us to express important measures of biological systems, such as probabilistic reachability, survivability, oscillations, switching times between attractor regions, and various others. Because of the infinite state space, available analysis techniques only apply to a very restricted subset of CSL properties. We present a full algorithm for model checking CSL for MPMs, and provide experimental evidence showing that our method is effective.

Fast Entropic Profiler: An Information Theoretic Approach for the Discovery of Patterns in Genomes.

Information theory has been used for quite some time in the area of computational biology. In this paper we present a pattern discovery method, named Fast Entropic Profiler, that is based on a local entropy function that captures the importance of a region with respect to the whole genome. The local entropy function has been introduced by Vinga and Almeida in , here we discuss and improve the original formulation. We provide a linear time and linear space algorithm called Fast Entropic Profiler ( FastEP), as opposed to the original quadratic implementation. Moreover we propose an alternative normalization that can be also efficiently implemented. We show that FastEP is suitable for large genomes and for the discovery of patterns with unbounded length. FastEP is available at http://www.dei.unipd.it/~ciompin/main/FastEP.html.

Combining Experiments and Simulations Using the Maximum Entropy Principle

A key component of computational biology is to compare the results of computer modelling with experimental measurements. Despite substantial progress in the models and algorithms used in many areas of computational biology, such comparisons sometimes reveal that the computations are not in quantitative agreement with experimental data. The principle of maximum entropy is a general procedure for constructing probability distributions in the light of new data, making it a natural tool in cases when an initial model provides results that are at odds with experiments. The number of maximum entropy applications in our field has grown steadily in recent years, in areas as diverse as sequence analysis, structural modelling, and neurobiology. In this Perspectives article, we give a broad introduction to the method, in an attempt to encourage its further adoption. The general procedure is explained in the context of a simple example, after which we proceed with a real-world application in the field of molecular simulations, where the maximum entropy procedure has recently provided new insight. Given the limited accuracy of force fields, macromolecular simulations sometimes produce results that are at not in complete and quantitative accordance with experiments. A common solution to this problem is to explicitly ensure agreement between the two by perturbing the potential energy function towards the experimental data. So far, a general consensus for how such perturbations should be implemented has been lacking. Three very recent papers have explored this problem using the maximum entropy approach, providing both new theoretical and practical insights to the problem. We highlight each of these contributions in turn and conclude with a discussion on remaining challenges.