Development ES
Exploitation ES
support ES
ES
182
Figure 3 – ES use in the learning process.
DET trainees who are just being familiarized with programming may by means of ES develop
independently applications where they may receive at once highly qualified maintenance on the work under
development.
Using of moduli of speech recognition programs plays the role of no little importance. The speech
recognition program allows achieving a higher level of management and fulfillment of the set tasks on
E
d
u
c
a
t
i
o
n
Data
IC
Сontents
Electronic books
Electronic Library
Student
IC
elective subjects
The educational process
…….
M
a
n
a
g
e
m
e
n
t
Corrective Actions
L e v e l 1
------
D
at
a
-
in
Make a
decision
Database-X2
Database1
Database2
Database3
Database-X1
Database4
…
Laboratory
Regulatory Documents
Modules Expert
Systems
L e v e l 3
L e v e l 2
L e v e l N
Analyze
Data
Results
Imple-
mentation
183
conduct of laboratory-and-research works while working remotely from the principal server. Thereby it is
possible to create additional opportunities, which simplify the process of data management and entry,
mastering of educational programs by the students with the use of distance learning technologies.
Integration of modern decisions of moduli for speech recognition programs to the distance education
systems provides for improvement of the educational portal.
Using of the speech recognition programs for DE test programs gives a number of advantages. Many
higher educational institutions put in practice the method of provision of information resources in the form of
text files or documents and knowledge control is implemented by the method of computer testing (in some
cases without the trainee personal identification in the on-line mode). In our opinion, such democratic
approach in higher education is inacceptable and rather fits for qualification improvement and etc. For
example, Educational Testing Service (ETS) company introduces highly-technological platforms of personal
identification for examination conduct for the purpose of providing an impeccable reputation in the world.
While ETS, the test development company itself, has introduced biometrical programs of voice identification
for the purpose of increasing the testing safety and also reliable and fair test conduct all over the world. As a
supplement to the existing complex safety system of the program for test participant authentication, the
proved safety method forms the basis of the recently launched safety measure [6]. Using of such the most
modern components improves the ability to detect attempts of receiving unfair advantage that is now the
common concern in the academic group of not only Kazakhstan but many open universities of the world [7].
REFERENCES
1. http://www.edu.gov.kz.
2. Makulov K., Otarbayev Zh. Significance of expert systems in distance educational technologies. Human
Forum. Barcelona. Spain. 2013.
3. И.К. Корнеев, В.В. Годин. Управление информационными ресурсами. – М.: ИНФРА-М, 2000. – 352 с.
4. http://wikipedia.org/
5. http://www.ruslion.ru/psyhology
6. http://www.mbastrategy.ua/content
7. Otarbayev Zh., Makulov K. Materials of international scientific-practical conference. University of SCO –
New horizons for distance education: Experience, practice and prospects. For the issues of pedagogic communication
in learning with the use of distance education technologies. Karaganda. Republic of Kazakhstan. 2013. – 222c.
8. Makulov K., Otarbayev Zh., Yagaliyeva B. Significance of Expert Systems in Distance Educational
Technologies // Materials of the VII International Research and Practice Conference Vol. II April 23h – 24th, 2014
Munich, Germany 2014. P.69-72
Макулов К.К., Отарбаев Ж., Ягалиева Б.Е.
Аннотация: В статье рассмотрены экспертные системы. Вопросы их применения в образовании с
дистанционными образовательными технологиями. Отражены основные недостатки и определение путей их
решения в целях создания модуля ЭС для системы дистанционных образовательных технологии.
Макулов К.К., Отарбаев Ж., Ягалиева Б.Е.
Аңдатпа: Бұл мақалада сараптық жүйе қарастырылған. Қашықтан оқыту технологиясының білім беруде
қолданылу мәселелері. Қашықтан оқыту технологиясы жүйесі үшін сараптық жүйе модулін құрудағы
шешімдерін анықтау жолдары мен негізгі кемшіліктері көрсетілген.
UDC004.89:004.4:681.5
Samigulina G.А., Samigulina Z.I.
Kazakh National Technical University after K.I. Satpaev,
Almaty, Republic of Kazakhstan
zarinasamigulina@mail.ru
INTELLECTUAL TECHNOLOGY OF IMMUNE NETWORK MODELING
FOR PREDICTION OF MEDICINES PROPERTIES
Abstract. Given article is devoted to development of the intellectual technology, computing algorithms and
programs for computer molecular design of the medical products with given properties (on an example barbiturate)
with usage immune net modeling. Classification of chemical substances on the prognostic groups is carried out.
Necessary requirements to the intellectual artificial immune system (AIS) for research of the relation between structure
and activity of the chemical compounds are considered. The system approach on the basis of association of the chemical
184
structural information processing methods with molecular modeling and the images recognition on the basis of AIS are
developed. The intellectual technology immune net forecasting of the barbiturates properties on the basis
multialgorithmic approach is offered.
Key words: artificial immune system, intellectual technology, prediction of dependence of the structure –
property, barbiturates, computer molecular design.
Introduction
Nowadays, the development of new medicines is a very important task. All over the world conducted
the researches on this problem. In the paper [1] authors used artificial neural network, which is a branch of
artificial Intelligence (AI). An analysis by neural networks improve the classification accuracy, data
quantification. In present study, an effort is being made to prepare the logical assembling of and reduce the
number of analogues necessary for correct classification of biological active compounds the various
advanced methods which will be circulating around the Artificial Neural Network. It is reported that drug
industries need the fast screening of chemical molecule to determine drug like properties in molecules.
Paper [2] describes the implementation of the Tabu Search (TS) algorithm in concert with the
Computer-Aided Molecular Design (CAMD) scheme. Although other optimization approaches have been
applied to CAMD with properties predicted using group contribution techniques, the TS algorithm
implemented with novel neighbor-generating operators and combined with property prediction via
connectivity index based correlations provides a powerful technique for generating lists of near-optimal
molecular candidates for a given application.
Paper [3] described peptide deformylase (PDF), which is essential in a variety of pathogenic bacteria
but it is not required for cytoplasmic protein synthesis in eukaryotes, which makes this enzyme an attractive
target for developing novel antibiotics. Authors designed a series of PDF inhibitors and predicted their
biological activities using molecular simulation methods. The binding conformations and binding affinities
of these inhibitors have been obtained using the flexible docking protocol FlexX. Calculations performed for
test compounds suggested that FlexXcan reproduce the binding conformation of the crystal structure. A
series of designed PDF inhibitors have been docked to the PDF model and the computed docking scores have
been used as a reference standard to evaluate the activities of these inhibitors.
Article [4] presents a dynamic ensemble neural network model for a pharmaceutical drug design
problem. By designing a drug, we mean to choose some variables of drug formulation (inputs), for obtaining
optimal characteristics of drug (outputs). To solve such a problem, authors propose a dynamic ensemble
neural network model and the performance is compared with several neural network architectures and
learning approaches. The idea is to build a dynamic ensemble neural network depicting the dependence
between inputs and outputs for the drug design problem. In paper [5] described a construction method and a
training procedure for a topology preserving neural network (TPNN) in order to model the sequence-activity
relation of peptides. The building blocks of a TPNN are single cells (neurons) which correspond one-to-one
to the amino acids of the peptide. The cells have adaptive internal weights and the local interactions between
cells govern the dynamics of the system and mimic the topology of the peptide chain. The TPNN can be
trained by gradient descent techniques, which rely on the efficient calculation of the gradient by back-
propagation.
Article [6] is devoted to prediction of anticancer/non-anticancer drugs. The quantitative structure-
activity relationship (QSAR) model developed discriminate anticancer/non-anticancer drugs using machine
learning techniques: artificial neural network (ANN) and support vector machine (SVM). The ANN used
here is a feed-forward neural network with a standard back-propagation training algorithm. These methods
were trained and tested on a non redundant data set of 180 drugs. The proposed model can be used for the
prediction of the anti-cancer activity of novel classes of compounds enabling a virtual screening of large
databases.
Authors of article [7] present self-organizing map or Kohonen network and counter propagation neural
network as powerful tools in quantitative structure property/activity relationship modeling. Two areas of
applications are discussed: estimation of toxic properties in environmental research and applications in drug
research.
Article [8] is devoted to describe the QSAR method. Antimicrobial peptides are ubiquitous in nature
where they play important roles in host defense and microbial control. QSARs method, which attempts to
correlate chemical structure to biological measurement, has shown promising results in the optimization and
discovery of peptide candidates.
In paper [9] authors reviewed the implementation of genetic algorithm (GA) in drug design QSAR
and specifically its performance in the optimization of robust mathematical models such as Bayesian-
185
regularized artificial neural networks (BRANNs) and support vector machines (SVMs) on different drug
design problems. The GA-optimized predictors were often more accurate and robust than previous published
models on the same data sets and explained more than 65% of data variances in validation experiments.
Also, in the process of the drug design there is used the relatively new biological approach [10] -
artificial immune system (AIS). In paper [11] present the first application of the artificial immune
recognition system (AIRS) to the recognition of the substrates of the multidrug resistance (MDR) ATP–
binding cassette (ABC) transporter permeability glycoprotein (P– glycoprotein, P– gp). We evaluated the
AIRS algorithm for a dataset of 201 chemicals. The classifiers were computed from 159 structural
descriptors from five classes, namely constitutional descriptors, topological indices, electro topological state
indices, quantum descriptors, and geometrical indices. The AIRS algorithm is controlled by eight user
defined parameters: affinity threshold scalar, clonal rate, permutation rate, number of nearest neighbors,
initial memory cell pool size, number of instances to compute the affinity threshold, stimulation threshold,
and total resources. The AIRS predictions are better than those of five of these algorithms (alternating
decision tree, Bayesian network, logistic regression with ridge estimator, random tree, and fast decision tree
learner), showing that P–gp substrates may be successfully recognized with AIRS. This article focuses on the
development of intellectual technology of immune network modeling and "structure-property" relations
prediction of unknown chemical compounds that can be considered as candidates for the creation of new
drugs.
The structure of the article is follow: section 2 is dedicated to the AIS approach description and to the
presentation of a mathematical model of a formal peptide, in Section 3 there are the objectives of the
analysis. Section 4 is dedicated to the immune network model construction based on the structure of the
compounds. There is also represented an algorithm for the prediction based on AIS and described the process
of immune network model optimization. Section 5 presents the results of the modeling. The article ends with
section 6 with the conclusion.
Description of the AIS approach
Under AIS there is understood an information technology, which use concepts of a theoretical
immunology for various applications, including the prediction of properties of new drug compounds. The
actual direction of AIS is the approach based on the mathematical implementation of the molecular
recognition mechanisms. The advantages of artificial immune system are: distribution, self-organization and
evolution, not much demanding of computer resources, the lack of centralized control, self-learning,
individual approach to the unique events.
This approach uses the term of a formal peptide [12], as a mathematical abstraction of the free energy
protein molecule from its spatial form, which is described in quaternion algebra.
A mathematical model of the formal peptide has the following form:
P=
(1)
where
0
n
- number of the elements;
k
k
U={
,
}, k=1 ,..., n
- multitude of the torsion angles, where
k
-
,
k
;
0
k
Q = {Q , Q }
- multitude of the unit quaternion, where quaternion
k
k
Q =Q {
,
}
k
k
and the resulting
quaternion FP
0
Q
are defined as their product
0
1
2
,
...
n
Q
Q Q
Q
; multitude of the coefficients:
j
V = {v }, = 1, 2, 3, 4, j
i
i
i
(2)
function
(without index), defined from the elements of the resulting quaternion
0
Q
by the following
quadratic form:
ij
i
j
j i
v
v q q
(3)
This mathematical model of the formal peptide is used in further researches during the immune
network model construction that describes the structure of a chemical compound by the descriptors.
Statement of the problem
The statement of the problem is formulated as follows: there is a need to develop an intellectual
system of forecasting the pharmacological activity of organic compounds on the basis of biological artificial
immune systems approach for the selection of promising chemical compounds -candidates for new drugs.
186
Immune network model construction and prediction of the new substances properties based on AIS
The implementation of the proposed intelligent prediction technology of "structure-property" relations
includes the following major steps:
1. The choice of chemicals for research.
2. Structure description and classification.
3. Pattern recognition based on AIS.
4. Analysis and evaluation of the properties predicting results of unknown compounds.
The Figure 1 shows a block diagram of prediction of the pharmacological activity of unknown
compounds based on immune network modeling.
Figure 1- The block diagram of drugs properties predicting based on AIS
The following algorithm is developed on the basis of AIS:
1. Development of a descriptors database for the compound structure description by the numerical
parameters.
2. Pre-processing: normalization, centering and data recovery.
3. Immune network model construction based on the descriptors (formal peptides - standards that are
regarded as antigens) that describe the structure of the test compound with known properties.
4. Optimal immune network model construction by informative descriptors providing (using factor
analysis algorithms and neural networks).
5. Education according to the immune network standards with the teacher and assessment.
6. Formation of the matrix - the images that will be considered as antibodies.
7. The pattern recognition [12] problem solution based on the singular value decomposition (SVD)
and finding the minimum energy between formal peptides (antibodies and antigens).
8. Evaluation of the power errors based on homologous proteins.
9. Selection for the further optimization algorithm usage of the immune network (principal
component analysis or neural network approach), which gives the minimum error of generalization.
10. Prediction of pharmacological activities of unknown compounds.
11. Selection of connections - drug candidates for further research.
Modeling results
Example. Let’s consider the following example of drugs properties predicting in the class of
barbiturates.
Barbiturates(barbiturates) - a group of drugs, barbituric acid derivatives (CONHCOCH2CONH),
providing a depressing effect on the central nervous system [13]. These drugs have a sedative, anticonvulsant
187
and narcotic effect. Various barbiturates have their lasting effects on the body. There are drugs with long-
term (barbital, phenobarbital, barbital sodium), medium (cyclobarbital, barbamil, etaminalsodium) and short
(geksobarbital), duration of action. For this example, information about the structures of chemical
compounds of barbiturates are taken from the books of E.W. Stuper, U. Bruegger and P. Dzhurs [14].
Experts identify two classes:
Class 1 - strongly acting barbiturates (sleep duration discussed).
Class 2 - weak acting barbiturates.
There is created the descriptors database (Table 1). Reduction of uninformative descriptors and
optimal immune network model construction is made by using factor analysis and principal component
analysis (PCA) [15].
Table 1- Descriptors database
№
D1
D2
D3
D4
...
D25
1
0.280
0.040
1.187
0.253
...
0.100
2
0.278
0.044
1.192
0.255
...
0.118
3
0.282
0.042
1.191
0.249
...
0.111
4
0.279
0.046
1.187
0.251
...
0.113
5
0.283
0.430
1.189
0.248
...
0.118
...
...
...
...
...
...
...
9
0.278
0.044
1.192
0.255
...
0.119
…
…
…
…
…
…
…
The procedure of factor analysisis carried out in the statistical treatment of the data package of SPSS.
Component Figure in the rotated space is shown in Figure 2.
Достарыңызбен бөлісу: |