Springer handbook of bio neuroinformatics

1
Understandin
1. Understanding Nature Through the Symbiosis
of Information Science, Bioinformatics
and Neuroinformatics
Nikola Kasabov
This chapter presents some background infor-
mation, methods and techniques of information
science, bio- and neuroinformatics in their sym-
biosis. It explains the rationale, the motivation and
the structure of the Handbook that reflects on this
symbiosis. For this chapter some text and figures
from [1.1] have been used. As introductory chapter,
it gives a brief overview of topics covered in the
Springer Handbook of Bio-/Neuroinformatics with
the emphasis on the symbiosis of the three areas of
science concerned: information science (informat-
ics); bioinformatics; neuroinformatics. The topics
presented and included in the handbook are far
for being an exhaustive coverage of the three ar-
eas. But, they show clearly that we can better
understand Nature only if we utilize the methods
of IS, BI and NI in their integration and interaction.
1.1 Nature as the Ultimate Inspiration
and Target for Science .......................... 1
1.2 Information Science (IS) ........................ 3
1.2.1 The Scope of IS............................. 3
1.2.2 Probability,
Entropy and Information............... 3
1.2.3 Statistical Information Methods ..... 4
1.2.4 Machine-Learning Methods........... 5
1.2.5 Knowledge Representation............ 6
1.3 Bioinformatics...................................... 7
1.3.1 Biology Background...................... 7
1.3.2 Data Analysis and Modeling
in Bioinformatics ......................... 8
1.4 Neuroinformatics.................................. 9
1.4.1 Human Brain and Nervous System.. 9
in Neuroinformatics ..................... 10
1.5 About the Handbook............................. 11
1.5.1 Bioinformatics ............................. 11
1.5.2 Neuroinformatics ......................... 11
1.6 Conclusion ........................................... 12
References .................................................. 12
1.1 Nature as the Ultimate Inspiration and Target for Science
Science aims at understanding Nature. Science meth-
ods are inspired by principles from Nature, too. The
beauty of our world is that, along with the fascinat-
ing geographical formations, it has Life as a variety of
biological species. Most importantly – it has the high-
est level of Life – the conscious brain. Nature and Life
have been the ultimate inspirations and targets for the
three important areas of science – information science
(IS, or also called informatics), bioinformatics (BI) and
neuroinformatics (NI).
IS deals with generic methods for information pro-
cessing. BI applies these methods to biological data and
also develops its own specific information processing
methods. NI applies the methods of IS on brain and
nervous system data and also develops its own specific
methods.
Nature evolves in time. The most obvious example
of an evolving process is Life. Life is defined in the
Concise Oxford English Dictionary:
a state of functional activity and continual change
peculiar to organized matter, and especially to the
portion of it constituting an animal or plant before
death; animate existence; being alive.
Continual change, along with certain stability, is what
characterizes Life. Modeling living systems requires
that the continuous changes are represented in the
PartA1

2 Part A Understanding Information Processes in Biological Systems
model, i. e. the model adapts in a life-long mode and
at the same time preserves some features and principles
that are characteristic to the process. The stability–
plasticity dilemma is a well-known principle of life.
In a living system, evolving processes are observed at
different levels (Fig. 1.1).
At the quantum level, particles are in a complex
evolving state all the time, being at several locations at
the same time, which is defined by probabilities. Gen-
eral evolving rules are defined by several principles,
such as entanglement, superposition, etc. [1.2–6]. The
discovery of the atomic structure in Physics and Chem-
istry revolutionized the understanding of the quantum
principles. Among the scientists who contributed to that
is Ernest Rutherford (1871–1937) (Fig. 1.2).
At a molecular level, RNA (ribonucleic acid) and
protein molecules, for example, evolve and interact in
a continuous way based on the DNA (deoxyribonucleic
acid) information and on the environment. The central
dogma of molecular biology constitutes a general evolv-
ing rule, but what are the specific rules for different
species and individuals? The area of science that deals
with the information processing and data manipulation
at this level is BI. At the cellular level (e.g. neuronal
cells) all the metabolic processes, the cell growing, cell
division etc., are evolving processes [1.7].
At the level of cell ensembles, or at a neural net-
work level, an ensemble of cells (neurons) operates
in a concert, defining the function of the ensemble
or the network through learning, for instance – per-
ception of sound, perception of an image or learning
languages [1.8].
In the human brain, complex dynamic interactions
between groups of neurons can be observed when cer-
tain cognitive functions are performed, e.g. speech and
language learning, visual pattern recognition, reason-
ing, and decision making [1.9–11].
At the level of population of individuals, species
evolve through evolution (Charles Darwin). A bio-
6. Evolutionary (population/generation) processes
5. Brain cognitive processes
4. System information processing (e.g. neural ensemble)
3. Information processing in a cell (neuron)
2. Molecular information processing (genes, proteins)
1. Quantum information processing
Fig. 1.1 Six levels of evolving processes in a higher level
living organism (after [1.1])
logical system evolves its structure and functionality
through both lifelong learning of an individual and the
evolution of populations of many such individuals. In
other words, an individual is a result of the evolu-
tion of many generations of populations, as well as
a result of its own developmental lifelong learning pro-
cesses [1.11,12].
Processes at different levels from Fig. 1.1 show gen-
eral characteristics, such as listed below.
1. Frequency: Frequency, denoted as F, is defined
as the number of a signal/event repetition over
a period of time T (seconds, minutes, centuries).
Some processes have stable frequencies, but others
change their frequencies over time. Different pro-
cesses from Fig. 1.1 are characterized by different
frequencies defined by their physical parameters.
Usually, a process is characterized by a spectrum of
frequencies. Different frequency spectrums charac-
terize brain oscillations (e.g. delta waves), speech
signals, image signals, quantum processes.
2. Energy: Energy is a major characteristic of any ob-
ject and organism. Albert Einstein’s most celebrated
energy formula defines energy E as depending on
the mass of the object m and the speed of light c as
E = mc2. Defining the energy in a living system is
more complicated. The energy of a protein, for ex-
ample, depends not only on the DNA sequence that
is translated into this protein, but on the 3-D shape
of the protein and on external factors.
3. Information: Information is a characteristic that can
be defined in different ways as is discussed in
Sect. 1.2 of this chapter.
4. Interaction: There are many interactions within each
of the six levels from Fig. 1.1 and across these
levels. Interactions are what make a living organ-
ism a complex one. Understanding them is also
a challenge for BI and NI. For example, there are
complex interactions between genes in a genome,
Fig. 1.2 Ernest Rutherford (1871–1937)
PartA1.1

Understanding Nature 1.2 Information Science (IS) 3
and between proteins and DNA. There are complex
interactions between the genes and the functioning
of each neuron, a neural network, and the whole
brain [1.13]. Abnormalities in some of these inter-
actions are known to have caused brain diseases and
many of them are unknown at present [1.14].
1.2 Information Science (IS)
Information science is the area of science that devel-
ops generic methods and systems for information and
knowledge processing regardless of the domain speci-
ficity of this information.
1.2.1 The Scope of IS
IS incorporates the following subject areas:
• Data collection and data communication (sensors
and networking)
• Information storage and retrieval (database systems)
• Methods for information processing (information
theory)
• Creating computer programs and information sys-
tems (software engineering and system develop-
ment)
• Acquisition, representing and processing knowl-
edge (knowledge engineering)
• Creating intelligent systems and machines (artificial
intelligence).
Generally speaking, data is raw entity – numbers, sym-
bols etc., e.g. 36.
Information is the labelled, understood, interpreted
data, e.g. the temperature of the human body is 36 ◦C.
Knowledge is the understanding of a human, the
way we do things, the interpretable in different situa-
tions general information, e.g., IF the human tempera-
ture is between 36 and 37 ◦C degrees, THEN the human
body is in a healthy state.
Some basic ways to represent data, information and
knowledge are presented in Sect. 1.2.2.
1.2.2 Probability, Entropy and Information
The formal theory of probability relies on the following
three axioms, where p(E) is the probability of an event
E to happen and p(¬E) is the probability of an event not
to happen. E1, E2, . . ., Ek is a set of mutually exclusive
events that form an universe U.
Axiom 1.1
0 ≤ p(E) ≤ 1.
Axiom 1.2
p(Ei) = 1, E1 ∪ E2 ∪. . .∪ Ek = U, U-problem space.
Corollary 1.1
p(E)+ p(¬E) = 1.
Axiom 1.3
p(E1 ∨ E2) = p(E1)+ p(E2), where E1 and E2 are mu-
tually exclusive events.
Probabilities are defined as:
• Theoretical – some rules are used to evaluate a prob-
ability of an event.
• Experimental – probabilities are learned from data
and experiments – throw dice 1000 times and
measure how many times the event getting 6 has
happened.
• Subjective – probabilities are based on common
sense human knowledge, such as defining that
the probability of getting 6 after throwing dice is
(1/6)th, without really throwing it many times.
A random variable x is characterized at any moment
of time by its uncertainty in terms of what value this
variable will take in the next moment – its entropy.
A measure of uncertainty h(xi) can be associated with
each random value xi of a random variable x, and the to-
tal uncertainty H(x), called entropy, measures our lack
of knowledge, the seeming disorder in the space of the
variable x
H(X) =
i=1,...,n
pih(xi) , (1.1)
where pi is the probability of the variable x taking the
value of xi.
The following axioms for the entropy H(x) apply:
• Monotonicity: if n > n are number of events (val-
ues) that a variable x can take, then Hn(x) > Hn (x),
so the more values x can take, the greater the en-
tropy.
PartA1.2

Fig. 1.3 Claude Shannon (1916–2011)
• Additivity: if x and y are independent random
variables, then the joint entropy H(x, y), meaning
H(x AND y), is equal to the sum of H(x) and H(y).
The following log function satisfies the two axioms.
h(xi) = log(1/pi) . (1.2)
If the log has a basis of 2, the uncertainty is measured
in [bit], and if it is the natural logarithm ln, then the
uncertainty is measured in [nat],
H(x) =
i=1,...,n
[pih(xi)] = −c
i=1,...,n
(pi log pi) ,
(1.3)
where c is a constant.
Based on the Shannon’s (Fig. 1.3) measure of uncer-
tainty – entropy, we can calculate an overall probability
for a successful prediction for all states of a ran-
dom variable x, or the predictability of the variable as
a whole
P(x) = 2−H(x)
. (1.4)
The maximum entropy is calculated when all the n val-
ues of the random variable x are equiprobable, i. e. they
have the same probability 1/n – a uniform probability
distribution
H(x) = −
i=1,...,n
pi log pi ≤ log n . (1.5)
Joint entropy between two random variables x and y (for
example, an input and an output variable in a system) is
defined by the formulas
H(x, y) = −
i=1,...,n
p(xi AND yj)
×log p(xi AND yj) , (1.6)
H(x, y) ≤ H(x)+ H(y) . (1.7)
Conditional entropy, i. e. measuring the uncertainty of
a variable y (output variable) after observing the value
of a variable x (input variable), is defined as
H(y|x) = −
i=1,...,n
p(xi, yj) log p(yj|xi) , (1.8)
0 ≤ H(y|x) ≤ H(y) . (1.9)
Entropy can be used as a measure of the information
associated with a random variable x, its uncertainty, and
its predictability.
The mutual information between two random vari-
ables, also simply called information, can be measured
as
I(y; x) = H(y)− H(y|x) . (1.10)
The process of online information entropy evaluation is
important as in a time series of events; after each event
has happened, the entropy changes and its value needs
to be re-evaluated.
Information models based on probability are:
• Bayesian classifiers
• Hidden Markov models (HMM).
A Bayesian classifier uses a conditional probability es-
timated to predict a class for a new data (1.11), which is
represented as the conditional probability between two
events C and A, known as the Bayes formula (Tamas
Bayes, 18th century)
p(A|C) = p(C|A)p(A)/p(C) . (1.11)
It follows from (1.11) that
p(A ∧C) = p(C ∧ A) = p(A|C)p(C)
= p(C|A)p(A) . (1.12)
Problems with the Bayesian learning models relate to
unknown prior probabilities and the requirement of
a large amount of data for more accurate probability cal-
culation. This is especially true for a chain of events
A, B, C, . . ., where probabilities p(C|A, B), . . . etc.
need to be evaluated. The latter problem is addressed
in techniques called hidden Markov models (HMM).
HMM [1.15] is a technique for modeling the tem-
poral structure of a time series, or, of a sequence of
events. It is a probabilistic pattern matching approach
which models a sequence of patterns as the output of
a random process. The HMM consists of an underlying
Markov chain
P[q(t +1)|q(t), q(t −1), q(t −2), . . ., q(t −n)]
≈ P[q(t +1)|q(t)] , (1.13)
where q(t) is state q sampled at a time t.
PartA1.2

Understanding Nature 1.2 Information Science (IS) 5
1.2.3 Statistical Information Methods
Correlation coefficients represent possible relationships
between variables. For every variable xi (i = 1, 2, . . .,
d1), its correlation coefficients Corr(xi, yj) with all
other variables yj ( j = 1, 2, . . ., d2), are calculated.
Equation (1.14) is to calculate the Pearson correlation
between two variables x and y based on n values for
each of them
Corr = SUMi
(xi −Mx)(yi −My)
(n −1)StdxStdy
, (1.14)
where Mx and My are the mean values of the two vari-
ables x and y, and Stdx and Stdy are their respective
standard deviations.
The t-test and the SNR (signal-to-noise ration)
methods evaluate how important a variable is to dis-
criminate samples belonging to different classes. For
a case of two class problem, a SNR ranking coefficient
for a variable x is calculated as an absolute difference
between the mean value M1x of the variable for class 1
and the mean M2x of this variable for class 2, divided
to the sum of the respective standard deviations
SNR x = abs
M1x −M2x
Std1x +Std2x
. (1.15)
A similar formula is used for the t-test
t-test x = abs
M1x −M2x
Std1x2/N1 +Std2x2/N2
, (1.16)
where N1 and N2 are the numbers of samples in class 1
and class 2, respectively.
Principal component analysis (PCA) aims at find-
ing a representation of a problem space X defined by
its variables X = {x1, x2, . . ., xn} into another orthog-
onal space having a smaller number of dimensions
defined by another set of variables Z = {z1, z2, . . ., zm},
such that every data vector x from the original space
is projected into a vector z of the new space, so that
the distance between different vectors in the original
space X is maximally preserved after their projection
into the new space Z.
Linear discriminant analysis (LDA) is a transfor-
mation of classification data from the original space
into a new space of LDA coefficients that has an ob-
jective function to preserve the distance between the
samples using also the class label to make them more
distinguishable between the classes.
Multiple Linear Regression Methods (MLR)
The purpose of MLR is to establish a quantitative re-
lationship between a group of p independent variables
(X) and a response y
y = XA+b , (1.17)
where p is the number of the independent variables, y is
an n ×1 vector of observations, X is an n × p matrix of
regressors; A is a p×1 vector of parameters, b is an
n ×1 vector of random disturbances. The solution to the
problem is a vector, A which estimates the unknown
vector of parameters.
The least squares solution is used so that the linear
regression formula or another model approximates the
data with the least root mean square error (RMSE) as
RMSE = SQRT SUM
i=1,2,...,n
(yi − yi)2
n
, (1.18)
where yi is the desired value from the data set corre-
sponding to an input vector xi; yi is the value obtained
through the regression formula for the same input vec-
tor xi and n is the number of the samples (vectors) in
the data set.
Another error measure is also used to evaluate the
performance of the regression and other models – a non-
dimensional error index (NDEI) – the RMSE divided to
the standard deviation of the data set
NDEI = RMSE/Std . (1.19)
1.2.4 Machine-Learning Methods
Machine learning is an area of IS concerned with the
creation of information models from data, with the
representation of knowledge, with the elucidation of in-
formation and knowledge from processes and objects.
Machine learning includes methods for feature selec-
tion, model creation, model validation and knowledge
extraction.
One of the widely used machine-learning method is
artificial neural networks (ANN) [1.9,11,16–22].
ANNs are computational models that mimic the ner-
vous system in its main function of adaptive learning.
An ANN consists of small processing units – artifi-
cial neurons, connected between each other. It has two
major functions – learning, which is the process of
presenting examples to the ANN and changing the con-
nection weights, and recall – the process of presenting
new examples to the trained ANN and examining its re-
action (output). The connection between the neurons are
analogized to the synaptic weights in the nervous sys-
tem. Most of the known ANNs learning algorithms are
influenced by a concept introduced by Hebb [1.23]. He
PartA1.2

Data
Knowledge
Information
Control
Adaption
Human
Environment
Pre-processing
Feature extraction
Labelling
Modelling
Learning
Knowledge discovery
Sound
Image
Video
Numbers
Sensors
Other sources
DNA
Brain signals
Stock market
Environmental and
social information
Fig. 1.4 From data to information and knowledge, and then back to information modeling (after [1.1])
proposed a model for unsupervised learning in which
the synaptic strength (weight) is increased if both the
source and the destination neurons become simultane-
ously activated. It is expressed as
wij(t +1) = wij(t)+coioj , (1.20)
where wij(t) is the weight of the connection between
the i-th and j-th neurons at the moment t, and oi and
oj are the output signals of neurons i and j at the same
moment t. The weight wij(t +1) is the adjusted weight
at the next time moment (t +1).
In general terms, a learning system {S, W, P, F,
L, J} is defined by its structure S, its parameter set
P, its variable (e.g., connections) weights W, its func-
tion F, its goal function J, and a learning procedure L.
The system learns if the system optimizes its structure
and its function F when observing events z1, z2, z3, . . .
from a problem space Z. Through a learning process,
the system improves its reaction to the observed events
and captures useful information that may be later repre-
sented as knowledge.
Another class of machine-learning methods are in-
spired by evolution of biology in Nature and is called
evolutionary computation (EC) [1.24, 25]. Here, learn-
ing is concerned with the performance not only of an
individual system, but of a population of systems that
improve their performance through generations. The
best individual system is expected to emerge and evolve
from such populations. EC methods, such as genetic
algorithms (GA) utilize ideas from Darwinism.
Another popular machine-learning method is called
support vector machines (SVM). It was first proposed
by Vapnik and his group at AT&T Bell Laborato-
ries [1.26]. For a typical learning task defined as
probability estimation of output values y depending on
input vectors x
P(x, y) = P(y|x)P(x) , (1.21)
a SVM classifier is used to build a decision function
fL : x → {−1, +1} (1.22)
based on a training set
fL = L(Strain) , (1.23)
where Strain = (x1, y1), (x2, y2), . . ., (xn, yn).
For SVM, the primary concern is determining an
optimal separating hyper-plane that gives a low gener-
alization error. Usually, the classification decision func-
tion in the linearly separable problem is represented by
fw,b = sign(w· x+b) . (1.24)
In SVM, this optimal separating hyperplane is de-
termined by giving the largest margin of separation
between vectors that belong to different classes. It bi-
sects the shortest line between the convex hulls of the
two classes.
1.2.5 Knowledge Representation
The ultimate goal of information processing is the cre-
ation of knowledge. The process of knowledge acquisi-
tion from Nature is a continuous process that will never
end. This knowledge is then used to understand Nature,
to preserve it, to model it and predict events. From data
to information and knowledge and then back, this is
what information science is concerned with (Fig. 1.4).
PartA1.2

Understanding Nature 1.3 Bioinformatics 7
Different types of knowledge can be used in the
machine learning systems, some of them being [1.1,21]:
• Propositional logic rules (Aristotle)
• First-order logic rules
• Fuzzy logic rules [1.27]
• Semantic maps
• Schemata
• Meta-rules
• Finite automata
• Higher-order logic.
Fuzzy logic is an extension of propositional logic. It
was first introduced by Zadeh in 1965 [1.27]. It deals
with fuzzy propositions that can have any truth value
between true (1) and false (0). Fuzzy propositions are
used to represent fuzzy rules that better represent human
knowledge [1.21,28], e.g.
IF gene expression of the gene G is High AND the
age of the person is Old THEN the risk of cancer is High,
where fuzzy membership functions (rather than Yes, or
No values) are used to represent the three propositions.
Modeling and knowledge can be [1.26,29]:
• Global: valid for the whole population of data
• Local: valid for clusters of data [1.1,30]
• Personalized: valid only for an individual [1.1, 26,
31].
1.3 Bioinformatics
1.3.1 Biology Background
Bioinformatics brings together several disciplines – mo-
lecular biology, genetics, microbiology, mathematics,
chemistry and biochemistry, physics, and of course –
informatics, with the aim of understanding life. The
theory of evolution through natural selection (Charles
Darwin, 1809–1882, Fig. 1.5) was a significant step to-
wards understanding species and life.
With the completion of the first draft of the human
genome and the genomes of some other species, the
task is now to be able to process this vast amount of
ever growing dynamic information and to discover new
knowledge.
The DNA (deoxyribonucleic acid) is a chemical
chain, present in the nucleus of each cell of an organism,
and it consists of ordered in a double helix pairs of small
chemical molecules (bases) which are: Adenine (A), cy-
tosine (C), guanine (G), and thymine (T), linked together
by sugar phosphate nucleic acid backbone (Fig. 1.6).
The central dogma of molecular biology states that
the DNA is transcribed into RNA, which is translated
into proteins.
The DNA contains millions of base pairs, but only
5% or so is used for the production of proteins, and
these are the segments from the DNA that contain
genes. Each gene is a sequence of base pairs that is used
in the cell to produce RNA and/or proteins. Genes have
a length of hundreds to thousands of bases.
The RNA (ribonucleic acid) has a similar structure
as the DNA, but here thymine (T) is substituted by
uracil (U). In the pre-RNA only segments that contain
genes are extracted from the DNA. Each gene consists
of two types of segments – exons, that are segments
translated into proteins, and introns – segments that are
considered redundant and do not take part in the protein
production. Removing the introns and ordering only the
exon parts of the genes in a sequence is called splicing
and this process results in the production of a messenger
RNA (or mRNA) sequences.
Fig. 1.5 Charles Darwin (1809–1882)
Fig. 1.6 DNA is organized as a double helix
PartA1.3

mRNAs are directly translated into proteins. Each
protein consists of a sequence of amino acids, each of
them defined as a base triplet, called a codon. From one
DNA sequence there are many copies of mRNA pro-
duced, the presence of a certain gene in all of them
defines the level of the gene expression in the cell and
can indicate what and how much of the corresponding
protein will be produced in the cell.
Genes are complex chemical structures and they
cause dynamic transformation of one substance into an-
other during the whole life of an individual, as well as
the life of the human population over many generations.
When genes are in action, the dynamics of the processes
in which a single gene is involved are very complex, as
this gene interacts with many other genes, proteins, and
the process is influenced by many environmental and
developmental factors.
Modeling these interactions, learning about them
and extracting knowledge, is a major goal for bioinfor-
matics.
Bioinformatics is concerned with the application of
the methods of information sciences for the analysis,
modeling and knowledge discovery of biological pro-
cesses in living organisms.
in Bioinformatics
There are five main phases of information process-
ing and problem solving in most of the bioinformatics
systems:
1. Data collection, e.g. collecting biological samples
and processing them.
2. Feature analysis and feature extraction – defin-
ing which features are more relevant and therefore
should be used when creating a model for a particu-
lar problem (e.g. classification, prediction, decision
making).
3. Modeling the problem – that consists of defining
inputs, outputs and type of the model (e.g., prob-
abilistic, rule-based, connectionist), training the
model and statistical verification.
4. Knowledge discovery in silico – new knowledge is
gained through the analysis of the modeling results
and the model itself.
5. Verifying the discovered knowledge in vitro and in
vivo – biological experiments in both laboratory
and in real life to confirm the discovered knowl-
edge.
When creating models of complex processes in
molecular biology the following issues must be consid-
ered:
• How to model complex interactions between genes
and proteins, between the genome and the environ-
ment.
• Both stability and repetitiveness are features that
need to be modelled, because genes are relatively
stable carriers of information.
• Dealing with uncertainty. For example, when mod-
eling gene expressions, there are many sources of
uncertainty, e.g., alternative splicing (a splicing pro-
cess of same RNAs resulting in different mRNAs);
mutation in genes caused by ionizing radiation (e.g.
x-rays); chemical contamination, replication errors,
viruses that insert genes into host cells, etc. Mutated
genes express differently and cause the production
of different proteins.
There are many problems in bioinformatics that require
their solutions through data analysis and modeling. Typ-
ical problems are:
• Discovering patterns from DNA and RNA se-
quences (e.g., promoters, binding sites, splice junc-
tions).
• Analysis of gene expression data and gene profile
creation.
• Protein discovery and protein function analysis.
• Modeling interaction networks between genes, pro-
teins and other molecules.
• System biology approach to modeling the whole dy-
namic process of biological functioning.
• Creating biological and medical prognostic and de-
cision support systems.
All the above tasks require different information meth-
ods, both generic – taken from IS, and specific – created
for the particular analysis and modeling of a specific
problem and type of biological data.
PartA1.3

Understanding Nature 1.4 Neuroinformatics 9
1.4 Neuroinformatics
1.4.1 Human Brain and Nervous System
The human brain can be viewed as a dynamic, evolv-
ing information-processing system – probably the most
complex one. Processing and analysis of information
recorded from brain and the nervous system activity,
modeling of perception, brain functions and cognitive
processes, aim at understanding the brain and creat-
ing brain-like intelligent systems. This is a subject of
neuroinformatics.
The brain evolves initially from stem cells. It
evolves its structure and functionality from an embryo
to a sophisticated biological information processing
system (Fig. 1.7).
As an embryo, the brain grows and develops based
on genetic information and nutritional environment.
The brain evolves its functional modules for vision, for
speech and language, for music and logic, and for many
cognitive tasks.
There are predeﬁned areas of the brain that are allo-
cated for language and visual information processing,
for example, but these areas may change during the
neuronal evolving processes. The paths of the signals
travelling and the information processes in the brain are
complex and different for different types of information.
Even at the age of three months, some functional ar-
eas are already formed. But identical embryos with the
same genetic information can develop in different ways
to reach the state of an adult brain. This is because of
the environment in which the brain evolves. Both the
genetic information (Nature) and the environment (Nur-
ture) are crucial factors. They determine the evolving
rules for the brain. The challenge is how to reveal these
rules and eventually use them in brain models? Are they
the same for every individual?
A signiﬁcant step in understanding the brain and the
nervous system was the discovery of the structure of the
neural system by Cajal (1852–1934, Figs. 1.8, 1.9).
A neuron, that receives signals (spikes) through its
dendrites and emits output signals through its axon is
connected to thousands other neurons through synapses.
The synaptic connections are subject to adaptation and
learning and represent the long term memory of the
brain.
Neurons can be of different types according to
their main functionality [1.11]. There are for exam-
ple: sensory neurons, motor neurons, local interneurons,
projection interneurons and neuroendocrine cells.
It is with the organization of the neurons in
ensembles that functional compartments emerge. Neu-
rosciences provide a very detailed picture of the
organization of the neural units in the functional com-
partments (functional systems). Each functional system
is formed by various brain regions that are responsible
to process different types of information. It is shown
that paths, which link different components of a func-
tional system are hierarchically organized.
It is mainly in the cerebral cortex where the cogni-
tion functions take place. Anatomically, cerebral cortex
is a thin, outer layer of the cerebral hemisphere with
thickness around 2–4 mm. Cerebral cortex is divided
into 4 lobes: frontal; parietal; temporal; occipital.
Several principles of the evolving structure, func-
tions and cognition of the brain are listed below:
• Redundancy, i. e. there are many redundant neurons
allocated to a single stimulus or a task; e.g. when
a word is heard, there are hundreds of thousands of
neurons that are immediately activated.
• Memory-based learning, i. e. the brain stores exem-
plars of facts that can be re-called at a later stage.
Some studies suggest that all human actions, in-
cluding learning and physical actions, are based on
memorized patterns.
• Learning is achieved through interaction of an
individual with the environment and with other in-
dividuals.
• Inner processes take place, e.g. information consol-
idation through sleep learning.
Primary motor
cortex
Motor association
cortex
Prefrontal
cortex
Speech center
Auditory cortex
Auditory association area
Visual association
area
Visual cortex
Primary somatosensory
cortex
Sensory association
cortex
Wernicke's area
Fig. 1.7 The brain evolves its structure and functionality through
genetic information and developmental learning (after [1.44])
PartA1.4

Fig. 1.8 Santiago Ramón y Cajal (1852–1934) (af-
ter [1.48])
Fig. 1.9 A drawing by Cajal of a neuronal circuitry (af-
ter [1.48])
• The learning process is continuous, evolving and
lifelong.
• Learning and memory of a three main types: short-
term (within hundreds of milliseconds) that is
manifested in the synapses, the neuronal membrane
potentials and the spiking activity of neurons; long
term (within minutes or days) that is manifested in
the synaptic weights; genetic (within months, years
or millions of years) that is manifested in the genes
and their expressions). These three types of memory
and learning are interacting in a dynamic way in the
brain.
• Through the process of evolving brain struc-
tures (neurons, connections) higher- level concepts
emerge; they are embodied in the structure and rep-
resent a level of abstraction.
in Neuroinformatics
The brain is the most complex information processing
machine. It processes data, information and knowledge
at different levels. Modeling the brain as an informa-
tion processing machine would have different results
depending on the goals of the models and the detail
with which the models represent the genetic, biological,
chemical, physical, physiological and psychological
rules and the laws that govern the functioning and be-
haviour of the brain.
Several levels of brain data analysis and modeling
can be distinguished.
Molecular/Genetic Level
At the genetic level the genome constitutes the input in-
formation, while the phenotype constitutes the output
result, which causes:
1. Changes in the neuronal synapses (learning)
2. Changes in the DNA and its gene expression [1.12].
Neurons from different parts of the brain, associ-
ated with different functions, such as memory, learning,
control, hearing and vision, function in a similar way
and their functioning is genetically defined. This prin-
ciple can be used as a unified approach to building
different neuronal models to perform different func-
tions, such as speech recognition, vision, learning and
evolving. The genes relevant to particular functions can
be represented as a set of parameters of a neuron. These
parameters define the way the neuron is functioning and
can be modified through feedback from the output of the
neuron.
Single Neuronal Level
There are many information models of neurons that
have been explored in the neural network theory (for
a review see [1.11]). Among them are:
1. Analytical models. An example is the Hodgkin–
Huxley model (1952) [1.32].
2. McCulloch and Pitts’ type models (1943) [1.33].
3. Spiking neuronal models (Maas, Gerstner, Kistler,
Izhikevich, Thorpe, Wysoski et al.) [1.29,34–43].
4. Neurogenetic models, where a gene regulatory net-
work is part of the neuronal model ([1.8,36,38,44]).
Neural Network (Ensemble) Level
Information is processed in ensembles of neurons that
form a functionally defined area, such as sensory
modalities [1.45]. The human brain deals mainly with
5 sensory modalities: vision, hearing, touch, taste and
smell. Each modality has different sensory receptors.
After the receptors perform the stimulus transduction,
the information is encoded through the excitation of
PartA1.4

Understanding Nature 1.5 About the Handbook 11
neural action potentials. The information is encoded us-
ing average of pulses or time interval between pulses.
This process seems to follow a common pattern for all
sensory modalities, however there are still many unan-
swered questions regarding the way the information is
encoded in the brain.
Cognitive Level
Information is processed in the whole brain as many in-
teracting modules. Many neuronal network modules are
connected together to model a complex brain structure
and learning algorithms [1.46]. To date, the most effec-
tive means available for brain activity measurement are
electroencephalography (EEG), magnetoencephalogra-
phy (MEG) and functional magnetic resonance imaging
(fMRI). Once the data from these measurement pro-
tocols has been transformed into an appropriate state
space representation, an attempt to model dynamic dif-
ferent brain functions can be made.
Modeling the entire brain is far from having been
achieved and it will take many years to achieve this
goal, but each step in this direction is an useful step to-
wards understanding the brain and towards the creation
of intelligent machines that will help people [1.47].
1.5 About the Handbook
This Springer Handbook includes 12 parts, 6 of them
covering topics from BI and 6 from NI. Each part in-
cludes chapters and each chapter introduces topics that
integrate BI and IS, or NI and IS, or BI, NI and IS
together.
1.5.1 Bioinformatics
Part A is about understanding information processes in
biological systems. It includes chapters that reveal the
information processing at cellular level, genomics level,
proteomics level, evolutionary molecular biology point
of view.
Part B covers the methods of molecular biology,
including: analysis of DNA sequences, analysis and dis-
covery of microRNA signatures, discovery regulatory
elements in RNA, protein data modeling and protein
structure discovery.
Part C presents different machine-learning meth-
ods for the analysis, modeling and knowledge discovery
from bioinformatics data. It includes chapters that
review the applications of different methods, such
as Bayesian classiﬁers and support vector machines
(SVM); case-based reasoning; hybrid clustering; fuzzy
logic; phylogenetic cladograms.
Part D presents more sophisticated methods for in-
tegrated, system biology analysis and modeling in BI,
including chapters on: inferring of interaction network
from Omics data; inferring gene transcription networks;
analysis of transcriptional regulations, inferring genetic
networks using differential evolution; pattern discov-
ery in protein-protein networks; visual representation
of molecular networks; a pipeline model for identify-
ing somatic mutations with examples from leukaemia
and colon cancer.
Part E presents databases and ontologies that con-
tain structured bioinformatics data to enable world-wide
research and study in bioinformatics. It includes chap-
ters on bioinformatics databases and bioinformatics
ontology systems.
Part F is about applications of bioinformatics in
medicine, heath and ecology. It includes chapters
on modeling cancer stem formation, epigenetics, im-
mune system control, nutrigenomics, nanomedicine,
personalized medicine, health informatics, ecological
informatics.
1.5.2 Neuroinformatics
Part G is about understanding information processes
in the brain and the nervous system. It includes chap-
ters on information processes at a lower, synaptic level,
spiking neural networks that represent and model infor-
mation processes at a neuronal ensemble level; brain
connectivity study based on fMRI data; information
processes at the level of the whole brain.
Part H introduces advanced signal processing meth-
ods for brain signal analysis and modeling. The
methods are applicable to study spatio-temporal spik-
ing activities of single neurons and neuronal ensembles
along with spiking activity of the whole cortex. This
part includes chapters on adaptive ﬁltering in kernel
spaces for spike train analysis, analysis and visuali-
sation of multiple spike trains, multivariate empirical
mode decomposition method for time-frequency anal-
ysis of EEG signals.
PartA1.5

Part I is concerned with modeling perception, sen-
sation and cognition. It includes chapters on modeling
vision, modeling gustatory system, perception and mo-
tor control modeling based on EEG with application for
brain-computer interfaces, spiking neural network and
neurogenetic systems for spatio- and spectro-temporal
brain data analysis and modeling, models of natural
language.
Part J presents neuroinformatics databases and sys-
tems to help brain data analysis and modeling. It
includes chapters on brain-gene ontology systems, neu-
roinformatics databases and world-wide organisations.
Applications of neuroinformatics methods for un-
derstanding and curing of brain diseases is presented
in Part K. It contains chapters on Alzheimer Dis-
ease genetic regulatory networks, integrating data and
prior knowledge for understanding Alzheimer Disease,
a system biology approach to modeling and understand-
ing Parkinson and Alzheimer Disease, modeling gene
dynamics in epilepsy, predicting outcome of stroke,
surface electromyography methods for nerve-muscle
system rehabilitation on the case study of stroke.
Nature inspired integrated information technolo-
gies, presented in the last Part L, combine different
principles from biology, brain and quantum level of
information processing (Fig. 1.1). It includes chapters
on brain-like robotics, interactive, developmental mul-
timodal robotic systems, quantum and biocomputing
integration, integrated brain-, gene- and quantum in-
spired computational intelligence.
1.6 Conclusion
This chapter presents a brief overview of topics covered
in the Springer Handbook of Bio-/Neuroinformatics
with the emphasis on the symbiosis of the three areas of
science concerned: information science (informatics);
bioinformatics; neuroinformatics. The topics presented
and included in the Handbook are far for being an ex-
haustive coverage of the three areas. But, they show
clearly that we can better understand Nature only if we
utilize the methods of IS, BI and NI in their integration
and interaction.
References
1.1 N. Kasabov: Evolving Connectionist Systems: The
Knowledge Engineering Approach (Springer, London
2007)
1.2 R.P. Feynman, R.B. Leighton, M. Sands: The Feyn-
man Lectures on Physics (Addison-Wesley, Redding
1965)
1.3 R. Penrose, The Emperor’s New Mind (Oxford Univ.
Press, Oxford 1989)
1.4 R. Penrose: Shadows of the Mind. A Search for the
Missing Science of Consciousness (Oxford Univ. Press,
Oxford 1994)
1.5 C.P. Williams, S.H. Clearwater: Explorations in Quan-
tum Computing (Springer, Berlin 1998)
1.6 M. Brooks: Quantum Computing and Communica-
tions (Springer, Berlin, Heidelberg 1999)
1.7 D.S. Dimitrov, I.A. Sidorov, N. Kasabov: Com-
putational biology. In: Handbook of Theoretical
and Computational Nanotechnology, Vol. 1, ed. by
M. Rieth, W. Sommers (American Scientiﬁc Publisher,
New York 2004), Chap. 21
1.8 N. Kasabov, L. Benuskova: Computational neuroge-
netics, Int. J. Theor. Comput. Nanosci. 1(1), 47–61
(2004)
1.9 F. Rosenblatt: Principles of Neurodynamics (Spartan
Books, New York 1962)
1.10 W. Freeman: Neurodynamics (Springer, London
2000)
1.11 M. Arbib (Ed.): The Handbook of Brain Theory and
Neural Networks (MIT, Cambridge 2003)
1.12 H. Chin, S. Moldin (Eds.): Methods in Genomic Neu-
roscience (CRC, Boca Raton 2001)
1.13 J.J. Hopﬁeld: Neural networks and physical systems
with emergent collective computational abilities,
Proc. Natl. Acad. Sci. USA 79, 2554–2558 (1982)
1.14 National Center for Biotechnology Information (US):
Genes and Disease [Internet] (NCBI, Bethesda 1998),
available online at http://www.ncbi.nlm.nih.gov/
books/NBK22183/
1.15 L.R. Rabiner: A tutorial on hidden Markov mod-
els and selected applications in speech recognition,
Proc. IEEE 77(2), 257–285 (1989)
1.16 S. Grossberg: On learning and energy – Entropy
dependence in recurrent and nonrecurrent signed
networks, J. Stat. Phys. 1, 319–350 (1969)
1.17 D.E. Rumelhart, G.E. Hinton, R.J. Williams (Eds.):
Learning internal representations by error prop-
PartA1

Understanding Nature References 13
agation. In: Parallel Distributed Processing: Ex-
plorations in the Microstructure of Cognition
(MIT/Bradford, Cambridge 1986)
1.18 T. Kohonen: Self-Organizing Maps (Springer, Berlin,
Heidelberg 1997)
1.19 S. Haykin: Neural Networks – A Comprehensive
Foundation (Prentice Hall, Engelwood Cliffs, 1994)
1.20 C. Bishop: Neural Networks for Pattern Recognition
(Oxford Univ. Press, Oxford 1995)
1.21 N. Kasabov: Foundations of Neural Networks, Fuzzy
Systems and Knowledge Engineering (MIT, Cam-
bridge 1996)
1.22 S. Amari, N. Kasabov: Brain-like Computing and In-
telligent Information Systems (Springer, New York
1998)
1.23 D. Hebb: The Organization of Behavior (Wiley, New
York 1949)
1.24 X. Yao: Evolutionary artiﬁcial neural networks, Int.
J. Neural Syst. 4(3), 203–222 (1993)
1.25 D.B. Fogel: Evolutionary Computation – Toward
a New Philosophy of Machine Intelligence (IEEE, New
York 1995)
1.26 V. Vapnik: Statistical Learning Theory (Wiley, New
York 1998)
1.27 Z.A. Zadeh: Fuzzy sets, Inf. Control 8, 338–353 (1965)
1.28 T. Yamakawa, H. Kusanagi, E. Uchino, T. Miki: A new
effective algorithm for neo fuzzy neuron model,
Proc. Fifth IFSA World Congress (IFSA, 1993) pp. 1017–
1020
1.29 N. Kasabov: Global, local and personalized mod-
eling and proﬁle discovery in Bioinformatics: An
integrated approach, Pattern Recognit. Lett. 28(6),
673–685 (2007)
1.30 M. Watts: A decade of Kasabov’s evolving connec-
tionist systems: A review, IEEE Trans. Syst. Man
Cybern. C 39(3), 253–269 (2009)
1.31 Q. Song, N. Kasabov: TWNFI – Transductive neural-
fuzzy inference system with weighted data normal-
ization and its application in medicine, IEEE Trans.
Fuzzy Syst. 19(10), 1591–1596 (2006)
1.32 A.L. Hodgkin, A.F. Huxley: A quantitative description
of membrane current and its application to conduc-
tion and excitation in nerve, J. Physiol. 117, 500–544
(1952)
1.33 W. McCullock, W. Pitts: A logical calculus of the ideas
immanent in nervous activity, Bull. Math. Biophys.
5, 115–133 (1943)
1.34 W. Gerstner: Time structure of the activity of neural
network models, Phys. Rev. 51, 738–758 (1995)
1.35 E. Izhikevich: Simple model of spiking neurons, IEEE
Trans. Neural Netw. 14(6), 1569–1572 (2003)
1.36 N. Kasabov, L. Benuskova, S. Wysoski: A Compu-
tational Neurogenetic Model of a Spiking Neuron,
IJCNN 2005 Conf. Proc., Vol. 1 (IEEE, New York 2005)
pp. 446–451
1.37 N. Kasabov, R. Schliebs, H. Kojima: Probabilis-
tic computational neurogenetic framework: From
modeling cognitive systems to Alzheimer’s dis-
ease, IEEE Trans. Auton. Ment. Dev. 3(4), 1–12
(2011)
1.38 N. Kasabov: To spike or not to spike: A probabilis-
tic spiking neuron model, Neural Netw. 23(1), 16–19
(2010)
1.39 G. Kistler, W. Gerstner: Spiking Neuron Models –
Single Neurons, Populations, Plasticity (Cambridge
Univ. Press, Cambridge 2002)
1.40 W. Maass, C.M. Bishop (Eds.): Pulsed Neural Networks
(MIT, Cambridge 1999)
1.41 S. Thorpe, A. Delorme, R. Van Rullen: Spike-based
strategies for rapid processing, Neural Netw. 14(6/7),
715–725 (2001)
1.42 S. Wysoski, L. Benuskova, N. Kasabov: Evolving spik-
ing neural networks for audiovisual information
processing, Neural Netw. 23(7), 819–835 (2010)
1.43 S. Guen, S. Rotter (Eds.): Analysis of Parallel Spike
Trains (Springer, New York 2010)
1.44 L. Benuskova, N. Kasabov: Computational Neuro-
Genetic Modeling (Springer, New York 2007)
1.45 E. Rolls, A. Treves: Neural Networks and Brain Func-
tion (Oxford Univ. Press, Oxford 1998)
1.46 J.G. Taylor: The Race for Consciousness (MIT, Cam-
bridge 1999)
1.47 D. Tan, A. Nijholt (Eds.): Brain-Computer Interfaces
(Springer, London 2010)
1.48 http://www.wikipedia.org (last accessed April 4
2012)
1.49 R. Koetter (Ed.): Neuroscience Databases: A Practical
Guide (Springer, Berlin, Heidelberg 2003)
PartA1

http://www.springer.com/978-3-642-30573-3

Springer handbook of bio neuroinformatics

More Related Content

Similar to Springer handbook of bio neuroinformatics (20)

More from Springer (20)

Recently uploaded (20)

Springer handbook of bio neuroinformatics