Inquiry (E-Journal - Faculty of Business and Administration, International University of Sarajevo)
Not a member yet
209 research outputs found
Sort by
Secondary Structure and Neighbor Preferences of Amino Acids
Get PDFThe mystery of the relation between amino acid sequences and folding of the proteins started to fascinate researchers starting from 1960’ies. When three-dimensional structures of globular proteins were first obtained by X-ray crystallography, there was no obvious relation foundbetween amino acid sequence and conformation. The ability of globular proteins refold from their denatured, time-random coils in the absence of other biological material, led some scientists to believe in that all the information for the native, biologically active conformation is contained within the amino acid sequence. In 1970’ies Anfinsen postulated that the native structure of a protein depends only on the amino acid sequence and on the conditions of solution, and not on the kinetic folding pathway. During that decade protein folding code was seen as a sum of many small interactions. But the key idea was that the primary sequence encoded secondary structures, which then encoded tertiary structures. In this article the claim that primary sequence encodes the secondary structure will be tested by the propensity of amino acids to helix-sheet-coil conformations
Protein Secondary Structure Prediction Using Super-chains in PDB
No full textThe completeness of the protein structures in the current Protein Data Bank (PDB) library for use in secondary structure prediction of unknown structure of protein is examined. To deal with this issue, randomly several 1000 protein chains batches are chosen from PDB. For each protein chain in the batch of PDB dataset that who contain the query protein chain as a subsequence are identified and named as a super-chain and prediction of the secondary structure of the query protein is performed by the use of the corresponding sub sequences of the secondary structure sequence of these chains. The technique is repeated for well known datasets such that CB513, FC699, 640, 25PDB, SCOP, and 1189 as well. It is seen that sequences of around 18% of proteins in the batch are present in other chains of PDB dataset. The average prediction accuracy of this method is found to be 80%. Therefore an unknown protein has a chance of 20% to have a super-chain in Protein Data Bank (PDB), and if a protein has a super-chain in the PDB database, there is a possibility that its secondary structure be predicted with around 80% accuracy
Modern Distribution Management System and Voltage VAR Control
Get PDFThis paper describes modern Distribution Management System (DMS) and Voltage/VAR Control (VVC) as one of its important components. Importance of DMS with respect to latest changes such as renewable energy sources, distribution generation, demand-respond is significant for the complete power system stability and control. In this paper VVC, as one of the most important applications in DMS, is explained and analyzed. VVC uses power system control equipment and calculates new optimal operational state. Typical VVC objective function is minimization of system power losses, violations of bus voltage limits, feeder capacity limits or combinations of these. Changes of controllable devices are presented through their injected current used in current iteration method for power flow. Test of Voltage/VAR control is performed on modified IEEE13 test network and results show that proper adjustments of OLTC transformers, capacitors and DG significantly reducepower losses while satisfying all operation constraints
Authorship Authentication for Twitter Messages Using Support Vector Machine
Get PDFWith the rapid growth of internet usage, authorship authentication of online messages became challenging research topic in the last decades. In this paper, we used a team of support vector machines to authenticate 5 Twitter authors’ messages. SVM is one of the commonly used and strong classification algorithms in authorship attribution problems. SVM maps the linearly non separable input data to a higher dimensional space by a hyperplane via radial base functions. Firstly using the training data, 10 hyperplanes that separate pair wise five authors training data are built. Then the expertise of these SVMs combined to classify the testing data into five classes. 20 tweets with 16 features from each author were used for evaluation. In spite of the randomly choice of the features, one of the author accuracy around 75% is achieved
A Literature Survey on Association Rule Mining Algorithms
No full textWith the development of database technology, the need for data mining arises. As a result, Association Rule Mining(ARM) has become a very hot topic in data mining. This paper presents definition and application areas of association rules. Furthermore, a comprehensive literature review on the existing algorithms of ARM is conducted with a special focus on the performance and application areas of the algorithms. These algorithms are in general classified into three main classes: (1) based on frequent itemset, (2) based on sequential pattern, and (3) based on structured pattern. The algorithms are developed to improve the accuracy and decrease the complexity, and execution time. However, it is hard to say that they do always succeed to optimize all these aspects simultaneously. Hence, there is still some space to develop more efficient algorithms for different data structures
Classification of Leaf Type Using Multilayer Perceptron, Naive Bayes and Support Vector Machine Classifiers
No full textMulticlass classification has always been challenging in the area of machine learning algorithms. Different publicly available software applications offer various learning algorithms’ implementations. This paper uses leaf dataset with 30 different plant species with simple leaf types prepared by Silva et al (2014), and classification is performed using Multilayer Perceptron, Naive Bayes and Support Vector Machine classifiers. Performance of classifiers is compared based on time needed for building the model and classification accuracy
Secondary Structure Segments are Much More Conserved than Primary Sequence Segments
Get PDFTo be biologically functional, all proteins must adopt specific folded three-dimensional structures. Some believes in that the genetic information for the protein specifies only the primary structure, the linear sequence of amino acids in the polypeptide backbone, and most purified proteins can spontaneously refold in vitro after being completely unfolded, so the three-dimensional structure must be determined by the primary structure (Creighton, 1990). How this occurs has come to be known as 'the protein folding problem'. As a part of the protein folding problem, the existence of similar substrings in diverse proteins is remarkable. Some scientist call it “conserved core” which echoes the claim that all proteins diversified from a common ancestor protein, and these similar pieces of the two or several proteins are the substrings that resisted the pressure of the evolution. Due to naturally-occurring (DNA fails to copy accurately) and external influences just like ultraviolet radiation, electromagnetic fields, atomic radiations, protein coding genes and proteins may undergo some changes by the time in response to mutations. The rate of these mutations is strongly correlated to the intensity of the environmental conditions, and it is not possible to estimate a constant rate just in the case of radioactive decay. Also there is no much evidence that the diversity of proteins relies on only these mutations. For this reason we prefer the term "similar substrings". In this paper we focused in the relation between primary and secondary structure mismatches of the substrings of length seventeen residues. We have seen that the mismatches in the corresponding secondary structure sequence substrings of the same length lags behind primary mismatches. We constructed a conditional probability landscape that resembles the conditional probability of a certain secondary substring mismatch given the primary substring mismatch. This landscape shows that even when 6-7 mismatches exist in two primary substrings of length 17 that belong to the two different proteins, the probability of full match of corresponding secondary structure substrings is remarkable. We downloaded primary and secondary sequences of all 303,524 proteins of the PDB protein databank. Eliminating the duplicates and proteins of residue length less than 30, we have got a non redundant database of 80,592. We developed a search algorithm FIND-SIM to find similar primary sequence substrings in a query protein and target proteins. Some examples of full secondary structure matches of short substrings corresponding to short primary structure substrings with high mismatches are given
Application Of Machine Learning In Healthcare: Analysis On MHEALTH Dataset
Get PDFThe healthcare services in developed and developing countries are critically important. The use of machine learning techniques in healthcare industry has a vital importance and increases rapidly. The corporations in healthcare sector need to take advantage of the machine learning techniques to obtain valuable data that could later be used to diagnose diseases at much earlier stages. In this study, a research is conducted with the purpose of discovering further use of the machine learning techniques in healthcare sector. Research was conducted by analyzing a well-established dataset called MHEALTH, comprising body motion and vital signs recordings for ten volunteers of diverse profile while performing 12 physical activities. Dataset was analyzed using certain classification algorithms such as Multilayer Perceptron and Support Vector Machine, then results from these algorithms were compared to determine the most utile algorithm for analyzing such dataset. Study aims to determine irregularities using data from body motion and vital signs of volunteers, then these findings can be used either to diagnose particular diseases before they occur and avoid them. Results can also be used to monitor movements of ill or elderly people and observe whether they are doing any prohibited movements that would lead them to injuries or further illnesses
Comparison of Different Machine Learning Algorithms for National Flags Classification
Get PDFEach country in the world has its own combination of colors, shapes and symbols on their flags. Some of them use an animal figure such as an eagle, some use an object like a boat; some nations prefer religion figures such as a crescent, or a cross. Some questions yet remain and need an answer. What are the factors that determine the flag of a nation? What factors are affecting the color or colors of a national flag? And what are the reasons for existence of symbols on some national flags?In this paper, we worked an analysis on national flags and factors that mostly affects the design of them. In order to find out these factors, we have used feature extraction method, after that we used different machine learning algorithms to predict religion and landmass of the country. We also showed correlations of certain components that are possible to exist on a national flag such as dominant color or colors on a flag, bars or stripes, normal and sacred symbols such as sun, stars, crosses, crescents, and triangles and, finally some specific icons like a boat or an animal figure.This study shows the associations of some characteristics of countries or different nationalities. There are many affected factors and there are very close correlations between these factors. It also includes the classification of national flag data using Multilayer Perceptron, CART and C4.5 algorithms and comparison of these techniques based on accuracy and performance for classification of national flag’s features
LU Factorization Algorithm with Minimum Degree Ordering in Power Distribution Network Problems
Get PDFPower systems computations for nowadays common large distributed systems typically involve the usage of very large sparse matrices, whose analysis and verification is very time and memory consuming. When blocked, sparse matrices can be processed much more efficiently, and this made blocked sparse matrices widely used in acquiring solutions for power system problems. The established sparse matrix storage and reordering techniques however do not fully utilize the existing computer architecture, thus search for efficient sparse system solution is ongoing. This paper presents adjustments of well-known LU factorization algorithm suitable for use in power distribution network applications. LU factorization algorithm processes data in blocks and uses minimum degree ordering to accelerate the computations