# Digital signal processing in the analysis of genomic sequences

**ABSTRACT** Digital Signal Processing (DSP) applications in Bioinformatics have received great attention in recent years, where new effective methods for genomic sequence analysis, such as the detection of coding regions, have been devel-oped. The use of DSP principles to analyze genomic sequences requires defining an adequate representation of the nucleo-tide bases by numerical values, converting the nucleotide sequences into time series. Once this has been done, all the mathematical tools usually employed in DSP are used in solving tasks such as identification of protein coding DNA re-gions, identification of reading frames, and others. In this article we present an overview of the most relevant applications of DSP algorithms in the analysis of genomic sequences, showing the main results obtained by using these techniques, analyzing their relative advantages and drawbacks, and providing relevant examples. We finally analyze some perspec-tives of DSP in Bioinformatics, considering recent research results on algebraic structures of the genetic code, which sug-gest other new DSP applications in this field, as well as the new field of Genomic Signal Processing.

**2**Bookmarks

**·**

**205**Views

- [Show abstract] [Hide abstract]

**ABSTRACT:**This paper presents a new technique for the detection of short exons in DNA sequences. In this method, we analyze four DNA structural properties, which include the DNA bending stiffness, disrupt energy, free energy, and propeller twist, using the autoregressive (AR) model. The linear prediction matrices for the four features are combined to find the same set of linear prediction coefficients, from which we estimate the spectrum of the DNA sequence and detect exons based on the 1/3 frequency component. To overcome the nonstationarity of DNA sequences, we use moving windows of different sizes in the AR model. Experiments on the human genome show that our multi-feature based method is superior in performance to existing exon detection algorithms.EURASIP Journal on Advances in Signal Processing. 01/2011; - SourceAvailable from: Vladimir Paar
##### Article: Direct mapping of symbolic DNA sequence into frequency domain in global repeat map algorithm.

[Show abstract] [Hide abstract]

**ABSTRACT:**The main feature of global repeat map (GRM) algorithm (www.hazu.hr/grm/software/win/grm2012.exe) is its ability to identify a broad variety of repeats of unbounded length that can be arbitrarily distant in sequences as large as human chromosomes. The efficacy is due to the use of complete set of a K-string ensemble which enables a new method of direct mapping of symbolic DNA sequence into frequency domain, with straightforward identification of repeats as peaks in GRM diagram. In this way, we obtain very fast, efficient and highly automatized repeat finding tool. The method is robust to substitutions and insertions/deletions, as well as to various complexities of the sequence pattern. We present several case studies of GRM use, in order to illustrate its capabilities: identification of α-satellite tandem repeats and higher order repeats (HORs), identification of Alu dispersed repeats and of Alu tandems, identification of Period 3 pattern in exons, implementation of 'magnifying glass' effect, identification of complex HOR pattern, identification of inter-tandem transitional dispersed repeat sequences and identification of long segmental duplications. GRM algorithm is convenient for use, in particular, in cases of large repeat units, of highly mutated and/or complex repeats, and of global repeat maps for large genomic sequences (chromosomes and genomes).Nucleic Acids Research 07/2012; 41(1):e47. · 8.81 Impact Factor - SourceAvailable from: Manish Kumar Gupta[Show abstract] [Hide abstract]

**ABSTRACT:**Biospectrogam is an open-source software for the spectral analysis of DNA and protein sequences. The software can fetch (from NCBI server), import and manage biological data. One can analyze the data using Digital Signal Processing (DSP) techniques since the software allows the user to convert the symbolic data into numerical data using 23 popular encodings and then apply popular transformations such as Fast Fourier Transform (FFT) etc. and export it. The ability of exporting (both encoding files and transform files) as a MATLAB .m file gives the user an option to apply variety of techniques of DSP. User can also do window analysis (both sliding in forward and backward directions and stagnant) with different size windows and search for meaningful spectral pattern with the help of exported MATLAB file in a dynamic manner by choosing time delay in the plot using Biospectrogram. Random encodings and user choice encoding allows software to search for many possibilities in spectral space. Availability: Biospectrogam is written in Java and is available to download freely from http://www.guptalab.org/biospectrogram. Software has been optimized to run on Windows, Mac OSX and Linux. User manual and you-tube (product demo) tutorial is also available on the website. We are in the process of acquiring open source license for it.10/2012;

Page 1

28 Current Bioinformatics, 2009, 4, 28-40

1574-8936/09 $55.00+.00 © 2009 Bentham Science Publishers Ltd.

Digital Signal Processing in the Analysis of Genomic Sequences

Juan V. Lorenzo-Ginori*,1, Aníbal Rodríguez-Fuentes1, Ricardo Grau Ábalo2 and Robersy Sánchez

Rodríguez3

1Centro de Estudios de Electrónica y Tecnologías de la Información, Facultad de Ingeniería Eléctrica, Universidad

Central “Marta Abreu” de Las Villas, Carretera a Camajuaní Km. 5 ?, 54830 Santa Clara, Villa Clara, Cuba; 2Centro

de Estudios de Informática, Facultad de Ingeniería Eléctrica, Universidad Central “Marta Abreu” de Las Villas, Car-

retera a Camajuaní Km. 5 ?, 54830 Santa Clara, Villa Clara, Cuba; 3Instituto Nacional de Investigaciones en Viandas

Tropicales, (INIVIT), Biotechnology Group, Santo Domingo, Villa Clara, Cuba

Abstract: Digital Signal Processing (DSP) applications in Bioinformatics have received great attention in recent years,

where new effective methods for genomic sequence analysis, such as the detection of coding regions, have been devel-

oped. The use of DSP principles to analyze genomic sequences requires defining an adequate representation of the nucleo-

tide bases by numerical values, converting the nucleotide sequences into time series. Once this has been done, all the

mathematical tools usually employed in DSP are used in solving tasks such as identification of protein coding DNA re-

gions, identification of reading frames, and others. In this article we present an overview of the most relevant applications

of DSP algorithms in the analysis of genomic sequences, showing the main results obtained by using these techniques,

analyzing their relative advantages and drawbacks, and providing relevant examples. We finally analyze some perspec-

tives of DSP in Bioinformatics, considering recent research results on algebraic structures of the genetic code, which sug-

gest other new DSP applications in this field, as well as the new field of Genomic Signal Processing.

Keywords: Digital Signal Processing, genomic sequences, coding regions.

INTRODUCTION

Digital Signal Processing (DSP) is an area of science and

engineering that has developed during the past 40 years as a

result of the constant evolution of computer science and

technology. DSP comprehends the representation, transfor-

mation and manipulation of digital signals as well as the in-

formation associated to them. In this context, signals are

usually physical magnitudes that vary in time or space, and

digital signals are those represented as sequences of num-

bers, as in the case of time series.

The discipline of DSP uses a set of mathematical tools to

analyze and process signals, among them can be mentioned

the Discrete Fourier Transform, the Z transform, Digital Fil-

ters, Parametric Models, the Wavelet Transform, Correlation

Functions and others. When considering the informational

content of signals, other concepts from Information Theory

such as entropy and mutual information are also used.

A key concept in DSP is the possibility of representing

the signals in the frequency domain making use of the Dis-

crete Fourier Transform. This representation leads to some

important signal properties that are not revealed in the time

domain, which are associated to their frequency spectrum.

In the case of the genomic sequences, these have been

represented mathematically by character strings of symbols

from a size-4 alphabet consisting of the letters A, T, G and

C, which represent each one of the nucleotide bases. In the

case of proteins, the alphabet size is 20, corresponding to the

*Address correspondence to this author at the Centro de Estudios de Elec-

trónica y Tecnologías de la Información, Facultad de Ingeniería Eléctrica,

Universidad Central “Marta Abreu” de Las Villas, Carretera a Camajuaní

Km. 5 ?, 54830 Santa Clara, Villa Clara, Cuba;

E-mail: juanl@uclv.edu.cu

possible amino acids. The possibility of finding a wide ap-

plication of DSP techniques to the analysis of genomic se-

quences arises when these are converted appropriately into

numerical sequences, for which several rules have been de-

veloped. Notice that genomic signals do not have time or

space as the independent variable, as occur with most physi-

cal signals.

This paper is organized in the following way. Firstly an

overview of the main DSP algorithms used in applications to

genomic sequence analysis is shown: digital filters, the Dis-

crete Fourier Transform (DFT), the Short-Time Fourier

Transform (STFT), parametric models (AR, MA, ARMA),

Wavelet Transform and the Information Theory concept of

entropy. Hidden Markov Models can be considered also as a

DSP tool, but this topic will not be covered, as there is a re-

cent comprehensive review article by De Fonzo et al. [1].

Then the numerical representation of genomic sequences is

presented. This allows the application of DSP tools to study

genomic sequences. After this, a review of the major appli-

cations of DSP to the analysis of genomic sequences is real-

ized, such as identification of protein coding DNA regions,

identification of reading frames, location of splice sites and

others. We finally review the perspectives of DSP in this

field, considering recent research results on algebraic struc-

tures of the genetic code and the new field of Genomic Sig-

nal Processing.

MAIN DSP ALGORITHMS EMPLOYED IN THE

ANALYSIS OF GENOMIC SEQUENCES

In this section a synthetic overview of the main DSP al-

gorithms that have been used in the analysis of genomic se-

quences is presented. There are excellent books on DSP the-

ory by Oppenheim and Schafer [2] and Proakis and Mano-

lakis [3].

Page 2

Digital Signal Processing in the Analysis of Genomic Sequences Current Bioinformatics, 2009, Vol. 4, No. 1 29

A) Digital Filters

A digital filter is a particular class of discrete system ca-

pable of realizing some transformation to an input discrete

numerical sequence. There are different classes of digital

filters according to the properties of their input-output rela-

tionships, as for example linear, nonlinear, time-invariant or

adaptive. The basic, frequency selective digital filters, are

linear and time-invariant (LTI) discrete systems.

Digital filters are characterized by numerical algorithms

that can be implemented in any class of digital processors. In

particular, LTI digital filters can pertain to one of two cate-

gories, according to the duration of their response to the im-

pulse, or Dirac delta function, when it is used as the input

signal: infinite (IIR) or finite (FIR) impulse response. The

input-output relationships for IIR digital filters are character-

ized and implemented algorithmically through a finite differ-

ence equation of the form

][][

00

knxbknya

M

k

?

=

k

N

k

?

=

k

?=?

, (1)

where x[n] and y[n] are the input and output numerical se-

quences respectively, ak and bk are numerical coefficients, n

is the sample index, and k is an integer delay with maximum

values N and M for the output and input sequences respec-

tively. On the other hand FIR digital filters are characterized

by a discrete convolution operation of the form

][][][

1

0

mnxmhny

N

m

?=?

?

=

(2)

In this equation, h[m] is the impulse response of the fil-

ter, which has a length of N samples. The bilateral Z trans-

form operator is defined as

?

??=

n

?

?

=

n

znxnxZ

][]}[{

(3)

where z is a complex variable. When this operator is applied

to equations (1) or (2), the system transfer function in the Z-

transform domain is obtained. The system transfer function

relates the input and output sequences x[n] and y[n], through

their respective Z transforms X[z] and Y[z]. The transfer

function has the general form

k

N

k

?

k

k

M

k

?

k

za

zb

zX

zY

zH

?

=

?

=

==

0

0

)(

)(

)(

(4)

The transfer function H(z) for this class of systems is a

ratio of polynomials in the complex variable z and has a

convergence region associated to it, which is closely related

to the positions of its poles in the complex Z plane. A prop-

erty of the transfer function of LTI systems is that the com-

plex exponential sequences of the form

][

nienx

?

=

where i is the imaginary unit, are eigenfunctions of these

systems, and this lead to the concept that these systems have

an associated frequency response, which can be obtained by

equating

z =

in equation (4), i.e.

?

ie

H(ei?)= H(z)]z=ei? (5)

The presence of the imaginary unit in the exponent im-

plies that H(ei?) is a complex function in the frequency do-

main, whose frequency response is usually expressed as a

magnitude response together with a phase, or angle response.

The system transfer function is periodic in ? (emphasizing

this periodicity is the reason for using ei?, instead of simply

?, as the argument of H), and it is usually plotted for its val-

ues in the main interval -???<?. An example of a sharp

resonance peak in the magnitude response of an IIR filter is

shown in Fig. (1), together with the corresponding phase

response. The sharp magnitude peak means a high selectivity

in frequency. The phase response of this filter is highly non-

linear (lower graph) and this nonlinearity tends to produce a

high signal distortion.

Fig. (1). Frequency response in magnitude and phase of an IIR

system exhibiting a sharp peak in the magnitude response.

A variety of digital filter design techniques allow to ob-

tain any desired magnitude response with frequency selectiv-

ity properties, whereas it is desired that the phase response

be a linear function of ?, in order to have low distortion.

According to the frequency interval (band) transmitted, the

magnitude of the basic ideal prototype filter frequency re-

sponses, can be lowpass, highpass, bandpass and bandstop.

A combination of these responses leads to a multiband filter.

The typical ideal frequency responses (in magnitude) of the

prototype filters are shown in Fig. (2). These ideal responses

can be only approximated in practical filters, where better

approximations in general are obtained by increasing the

order of H(z), which means a higher computational complex-

ity of the digital filters.

Constant magnitude response together with perfect line-

arity in the phase response is the condition for signal trans-

Page 3

30 Current Bioinformatics, 2009, Vol. 4, No. 1 Lorenzo-Ginori et al.

mission without distortion through a filter in the desired fre-

quency band. IIR digital filters have in general a nonlinear

phase response, that depends on the design method em-

ployed. On the other hand, a property of FIR digital filters is

that they can exhibit a perfect linear phase response under

certain conditions of symmetry in their impulse response.

This has been a motivation for the use of digital FIR filters in

many applications.

Fig. (2). Frequency response in magnitude for the prototype ideal

filters: lowpass, highpass, bandpass and bandstop.

B) Discrete Fourier Transform

The Discrete Fourier Transform is a mathematical

operation that transforms one discrete, limited (finite) N

duration function into another function, according to

?

=

n

?

?

=

1

0

2

N

][][

N

nki

enxkX

?

, 0? n, k ?N-1 (6)

The function X[k] is the Discrete Fourier Transform

(DFT) of the sequence x[n] and constitutes the frequency

domain representation of x[n], which is usually (or

conventionally considered) a function in the time domain.

The Discrete Fourier Transform only evaluates the frequency

components required to reconstruct the finite segment of the

sequence that was analyzed. In general, the DFT is a

function in the complex domain as a result of the complex

exponential in the right side of equation (6), and for the

particular case of real sequences, it will be a sequence of

complex numbers of the same length as x[n]. The DFT is

usually represented in terms of the corresponding magnitude

and phase functions that constitute the frequency spectrum of

the sequence x[n].

The Discrete Fourier transform is a very useful tool, be-

cause it can reveal periodicities in the input data as well as

the relative intensities of these periodic components. An ex-

ample of the magnitude and phase graphs of the 64-points

DFT for a sum of two pure sinusoids at discrete frequencies

14/2?

and

15/4?

is shown in Fig. (2). Each discrete value

of the DFT is usually called a DFT coefficient.

The DFT, however, suffer from three important draw-

backs as a tool for spectral analysis: a) Spectral leakage,

which means the presence of energy in zones where the

spectrum should be zero (this is clearly seen in Fig. (3): two

pure frequencies are analyzed while many nonzero samples

are obtained in the spectrum at other frequencies); b) the

frequency response of the DFT coefficients is not constant

with frequency (“picket-fence” effect), and c) the spectral

resolution, or ability to separate frequency lines that are

close in frequency, depends inversely upon the length of the

sequence in the time domain. This means that the DFT can-

not distinguish appropriately close spectral components for

time signals of short duration. Multiplying the time signals

by special weighting functions called windows, and control-

ling the signal length, can help in overcoming these limita-

tions in some extent.

Fig. (3). Example of DFT frequency spectrum (magnitude and

phase) for two sinusoids closely spaced in frequency. Frequency

axis is normalized to fs/N, where fs is the sampling frequency and N

the number of samples in the sequence (64 in this example).

Using the DFT for spectral analysis of random signals (or

stochastic processes) require certain considerations to obtain

a statistically valid result.

For stationary random signals, a commonly employed

procedure to obtain a power spectral density (PSD) function

in the frequency domain is the Welch’s modified perio-

dograms method. The PSD function is obtained in this case

by calculating the mean value of the squared DFT coeffi-

cients at each frequency value, for adjacent and usually over-

lapping windowed signal segments. The measure obtained in

this way is a consistent estimate of the power spectrum. A

typical spectrum obtained by the Welch’s method, for a pure

sinusoid embedded in white Gaussian noise, is shown in Fig.

(4). Notice the peak that corresponds to the sinusoid, whose

magnitude is significantly greater than the noisy background.

Page 4

Digital Signal Processing in the Analysis of Genomic Sequences Current Bioinformatics, 2009, Vol. 4, No. 1 31

Fig. (4). An example of PSD spectrum obtained through Welch’s

method, for a sinusoid embedded in white, Gaussian noise.

In the case of non-stationary signals, The Short Time

Fourier Transform (STFT) is an algorithm frequently used

for the DFT-based spectral analysis. In the STFT, the time

signal is divided into short segments (usually overlapped)

and a DFT is calculated for each one of these segments. A

three dimensional graph called spectrogram is obtained by

plotting the squared magnitude of the DFT coefficients as a

function of time. This squared magnitude is usually repre-

sented by the brightness of the graph, as shown in Fig. (5).

Fig. (5). Spectrogram of a harmonic signal whose frequency varies

linearly with time (“linear chirp”).

An important special case of the STFT is the Gabor

Transform, in which a Gaussian weighting window is ap-

plied to the analyzed time sequence. This procedure allows

obtaining a better simultaneous resolution in time and fre-

quency.

C) Spectral Analysis Using Parametric Models

Parametric spectral analysis is a method that can be used

in many cases with some advantages over the non-parametric

methods. Its advantages rely in that it is possible to obtain a

parametric description of the second-order statistics of a ran-

dom sequence, by assuming a certain production model for

it. A comprehensive analysis of such methods is given in

Stoica and Moses [4].

Spectral analysis using parametric methods does not suf-

fer from the limitations in spectral resolution that character-

ize the DFT-based methods, because they do not imply a

windowing (segment selection) process.

The mathematical expression of the PSD function of a

random sequence is described in this case in terms of the

model parameters, and the variance of a white (constant

PSD) random noise process used as the input signal of the

model. In consequence, the values to be computed in this

method are the parameters of the model and the variance of

the input process.

The general expression for the transfer function of the

model in parametric spectral analysis is analogous to that of

a digital filter as shown in equation (3), which is expressed

as the ratio of polynomials in the complex variable z

k

p

k

?

k

k

q

k

?

k

za

zb

zA

zB

zH

?

=

?

=

+

==

1

0

1

)(

)(

)(

(7)

to which corresponds the equation in finite differences

][][][

01

knwbknxanx

q

k

?

=

k

p

k

?

=

k

?+??=

(8)

in which w[n] is the input sequence and the observed data

x[n] represent the model’s output. Equations (7) and (8) are

related through the Z transform operator shown in equation

(3). The PSD function is obtained from (7) using (5) to ob-

tain the model’s frequency response, and is given by

?xx(?)= H(ei?)

2?ww(?) (9)

In equation (9) H(ei?) is the frequency response of the

model, while ?ww and ?xx are respectively the PSD functions

of the corresponding input and output signals. For a white-

noise input,

2

w

2

)()(

i

xx

eH

??

?

=?

(10)

where

2

w

? is the input noise variance.

According to the characteristics of the PSD for the ana-

lyzed random sequence there are three types of parametric

models:

•

Autoregressive (AR) models, corresponding to the

particular case {}

0

=

for k > 0, resulting in an all-

pole transfer function.

kb

•

Moving average (MA) models, which correspond

to{}

0

=

, resulting in an all-zero transfer function.

k a

•

Autoregressive, moving average (ARMA) models,

which is the general case in which there are poles and

zeros in the model’s transfer function.

Page 5

32 Current Bioinformatics, 2009, Vol. 4, No. 1 Lorenzo-Ginori et al.

There is equivalence between the three types of models if

the order is selected appropriately, i. e., a process which is

inherently AR of a certain order, can be described by an MA

model of higher order. However, AR models are more used

because of the relative simplicity in calculating the model’s

parameters through the Yule-Walker equations. Fig. (6)

shows the PSD curve for a typical AR spectrum.

Fig. (6). A typical PSD function obtained for an AR model, exhibit-

ing two peaks corresponding to two pairs of complex conjugate

poles in the model’s transfer function.

D) Discrete Wavelet Transform

The Discrete Wavelet Transform (DWT) is a mathemati-

cal tool that can be used very effectively for non-stationary

signal analysis. There is a great amount of literature on

DWT, see for example Burrus et al. [5].

In DWT analysis, a signal x(t) can be described through a

linear decomposition as

??

=

)()(

,,

tatx

kj

kj

kj?

(11)

In this equation j,k ? ? are integer indexes, aj,k are the

wavelet coefficients of the expansion, and ?j,k is a set of

wavelet functions in t. Notice that the wavelet coefficients

j a,constitute a discrete set, and that the coefficient’s values

are calculated according to

k

dtttxttxa

kjkjkj

?

??

+?

>==<

)()

?

()()

?

(

,,,

(12)

The DWT obtains the decomposition of the signal x[n]

into a set of orthonormal wavelets and their associated scal-

ing functions ?j,k that constitute a wavelet basis. These func-

tions can belong to different wavelet families that are ex-

pressed by the functions ?j,k which can be generated by dila-

tions and translations of a basic (“mother”) wavelet. These

dilations and translations are discrete, and the indexes j and k

are respectively related to these processes, that can be ex-

pressed as

()

ktt

jj

kj

?=

??

22)(

2/

,

??

, j,k? ? (13)

In Eq. (13) the functions ?j,k are dilated in a dyadic form

(in powers of two), when varying the values of the index j,

and in analogous way translated when varying the index k. In

this process, translation is associated with time resolution,

and dilation provides scaling, a concept closely related here

to frequency resolution.

Wavelet functions must satisfy the conditions

0)(lim

?

t

,

=

?

t

ji

?

(14)

and

?

??

?

= 0)(

,

dtt

ji

?

. (15)

In these conditions, (14) implies decay, and (15) implies

oscillations like a wave function. Fig. (7) shows examples of

wavelets functions that are well described in the literature.

Fig. (7). Examples of wavelets: (a) Daubechies Db10, (b) Coiflet

Coif5.

The DWT, for which an algorithm called Fast Wavelet

Transforms (FWT) allows a very efficient calculation, plays

currently a central role in many DSP applications. The result

of the DWT is a multi-resolution decomposition, in which at

each level the signal is decomposed in “approximation” and

“detail” coefficients. This decomposition is realized through

a process that is equivalent to lowpass and highpass filtering

for the approximation and for the details respectively, using

special digital filters called “Quadrature Mirror Filters”

(QMF.) There are two types of QMF filters: the lowpass

scaling filter h, and the highpass wavelet filter g. The g filter

is equivalent to the h filter reversed in time and alternating

the signs of its coefficients. DWT decompositions can be

depicted by a tree structure as shown in Fig. (8), where ap-

proximation and detail coefficients are represented. Each one

of the J decomposition levels corresponds to a certain dila-

tion j, whereas the index k determines the corresponding

translations. The DWT can be also extended to non-

orthogonal decompositions.

Page 6

Digital Signal Processing in the Analysis of Genomic Sequences Current Bioinformatics, 2009, Vol. 4, No. 1 33

Fig. (8). Approximation and Detail coefficients in a tree structure

for a DWT three-level decomposition. S is the original signal, cDi

and cAi stand respectively for detail and approximation coefficients

at level i.

E) Entropy Measures

Entropy measures are another example of a signal

processing concept that has been used in genomic sequence

analysis.

The concept of entropy is used in signal analysis as a

measure of randomness. The first definition of the entropy of

a discrete information source (producing a discrete sequence)

was introduced by Shannon [6] as

?

=

i

?=

N

ii

ppXH

1

log)(

(16)

where pi are the probabilities of the set of values that can

take the sequence X, {x1, x2, ... ,xn}.

Another definition frequently used is the Rényi entropy

[7], given by

?

=

i

?

=

n

ipXH

1

log

1

1

)(

?

?

?

(17)

Here H?(X) is the Rényi entropy of order ?, where

? ?0, and {pi} are the signal probabilities as defined before.

F) Final Remarks

Although in this section the more frequently used DSP

techniques were overviewed, it is important to notice that

there are other various important techniques that in some

cases have been used in the Bioinformatics field, such as

different transforms (Cosine, Sine, Walsh-Hadamard, Hil-

bert), fractal analysis, and others.

NUMERICAL REPRESENTATION OF GENOMIC

SEQUENCES

The first approach to convert genomic information in

numerical sequences was given by Voss [8] with the defini-

tion of the indicator sequences, defined as binary sequences

for each base, where 1 at position k indicates the presence of

the base at that position, and 0 its absence. For example,

given the DNA sequence

ACTTAGCTACAGA…

The binary indicator sequences X for each base A, T, C

and G are respectively:

XA[k] = 1000100010101…

XT[k] = 0011000100000…

XC[k] = 0100001001000…

XG[k] = 0000010000010…

The main advantages of the indicator sequences are their

simplicity, and the fact that they can provide a four-

dimensional representation of the frequency spectrum of a

character string, by means of computing the DFT of each

one of the indicator sequences. This dimensionality can be

reduced to three through the Z curves [9, 10] and the tetrahe-

dron [11] methods.

(18)

Another relevant numerical representation of genomic

sequences is a mapping in which a complex number is as-

signed to each base of the nucleotide sequence. In this case,

these complex numbers are appropriately selected to provide

useful properties of the numerical sequences. One of such

properties is obtained by assigning complex conjugate com-

plex numbers to the base pairs A, T and C, G. In this case all

palindromes will have conjugate symmetric numerical se-

quences. This lead to the generalized linear phase described

by Anastassiou [12]. A simple example of such mapping,

used in this reference is

1 ,1 ,1 ,1

jgjcjtja

+?=??=?=+=

(19)

where a, t, c and g are the numbers assigned respectively to

the bases A, T, C and G.

A more complete mapping that gives the representation

of all IUPAC nucleotide classes comprising single nucleo-

tides, doublets, triplets and quadruplets is given by Cristea et

al. in [13] and applied in [14] to analyze the variability of

pathogens’ genomes.

Other relevant criteria to select the numerical values to

represent genomic sequences are discussed by Akhtar et al.

[15]: equal magnitudes, equidistance, compactness of the

representation and easiness to use various mathematical

tools. Other examples of representations that have been used

are

3 , 2 , 1

=

, 0

===

gact

lois field assignment, and

in [16], which correspond to a Ga-

5 . 0

?

, 5 . 0 , 5 . 1

?

, 5 . 1

====

gcta

used in [15]. Notice that the latter shows the complementary

property, in the same way as in the complex assignment (19).

Rushdi and Tuqan [17] proposed a generic matrix based

framework that comprises most of the mappings reported in

the literature as special cases and can allow a number of po-

tential new mappings.

A representation of genomic sequences by means of qua-

ternions was introduced by Brodzik and Peters in [18], which

allows using the quaternionic Fourier Transform for pattern

detection in DNA sequences.

Page 7

34 Current Bioinformatics, 2009, Vol. 4, No. 1 Lorenzo-Ginori et al.

A relationship between the numerical assignment to the

nucleotides and to the amino acids has been established

through FIR digital filtering in [12].

APPLICATIONS OF DSP IN THE ANALYSIS GE-

NOMIC SEQUENCES

Digital Signal Processing applications to Bioinformatics

started in recent years in which great attention was put to the

problem of genomic sequence analysis. Fig. (9) depicts a

protein-coding DNA region and, in particular, a gene from

an eukaryotic genome, indicating the introns and exons and

the points where the gene begins (start codon), its end (stop

codon), donor splice sites (transition from an exon to an in-

tron), donor splice sites (transition from intron to exon) and a

CpG island (a region rich in CG pairs that may promote gene

function). Detecting all these places in a genomic sequence

is a source of application for DSP techniques.

One of the main motivations to introduce DSP in this

field was the find of hidden periodicities or oscillating pat-

terns in the genomic sequences, which were described by

Trifonov in [19] as 3, 10.5, 200 and 400-base periodicities.

Among them, the three-base periodicity was found to be a

characteristic of the protein-coding regions in both prokary-

otic and eukaryotic sequences.

The 3-periodicity is explained in more detail by Tuqan

and Rushdi [20] as related to the codon bias. Consider a ge-

nomic sequence analyzed through a rectangular window with

three-base length, that is displaced along the entire sequence

in three-base length intervals. The relative number of occur-

rences of base l in the kth (k=0, 1, 2) position of the codon in

the specific window positions, reveals that there is an unbal-

ance of the abundance of base l in codon position k with re-

spect to the average frequency of occurrence of base l in the

three possible codon positions. This phenomenon is reflected

in the frequency spectrum of the DNA sequence as a spectral

line exactly at N/3 in the DFT, N being the DFT length.

Another contribution to explain the three-base periodicity

was made by Sánchez and López-Villaseñor [21] through the

concept of same-phase triplet clustering, a condition in

which a triplet appears several times in one phase with no

interruptions by the two other possible phases.

Detection of Protein-Coding Regions Through Spectral

Analysis and the 3-Periodicity Property

A number of authors have devised algorithms to detect

protein coding regions in genomic sequences by finding re-

gions exhibiting a three-periodicity. Vaidyanathan and Yoon

[22] applied to the indicator sequences an anti-notch IIR

digital filter with a sharp narrow band centred at ?0 = 2?/3,

with the purpose of detecting the period 3 component. They

showed also lattice and multistage implementations, as well

as an equivalent DFT approach to this problem. The concept

that DNA sequences have an 1/f power spectrum that can be

considered as a noisy background, is used to argue that the

window length used to calculate the DFT should be long

enough, typically a few hundreds bp as 351, to a few thou-

sands, in order that the 3-periodicity dominates the noise

background. A typical result is given in Fig. (10), where

comparison to a threshold is usually employed to determine

the detected regions.

Fig. (10). Detection of 3-periodicity regions using DSP. Typical

plot in which noticeable peaks correspond to coding regions.

Another digital filtering approach, the polyphase Filtered

DNA spectrum, was presented by Tuqan and Rushdi [23].

Fox and Carreira [24] introduced a method in which only

one digital filter operation is required, followed by a quad-

ratic windowing operation which produces a signal that has

almost zero energy in the non-coding regions, improving the

effectiveness of the method.

Fig. (9). Diagram of a protein-coding DNA region and of a gene from an eukaryotic DNA, showing different characteristic points whose

detection is a source of applications of DSP techniques.

Page 8

Digital Signal Processing in the Analysis of Genomic Sequences Current Bioinformatics, 2009, Vol. 4, No. 1 35

The DFT approach to find the 3-periodicity regions in

genomic sequences has been used by various authors.

Afreixo et al. [25] analyze several methods for the Fourier

analysis of symbolic data oriented to DNA sequences, con-

sidering different approaches as the indicator sequences,

vector and symbolic correlation sequences and spectral enve-

lope.

Tiwari et al. [26] presented an early study of the applica-

tion of DFT analysis for gene prediction, where an experi-

mental study for a variety of genomic sequences was per-

formed. Another early example can be found in Yan et al.

[27], based in the format of the Z curve. Anastassiou [12]

used the DFT and the STFT spectrograms to analyze the

indicator sequences and introduced an optimized spectral

content measure to improve the discriminating properties of

the method. Datta et al. [28] used the DFT to find the 3-

periodicity regions and formalized mathematically some

properties of the DNA sequences. A fast DFT based gene

prediction algorithm and a DFT based splicing algorithms

are presented by these authors in [29, 30].

Isaac et al. [31] showed FTG, a web server to predict

genes based on DFT techniques, which allows rapid visuali-

zation by providing an output in GIF format. Stoffer et al.

[32] presented a study on the local spectral envelope used

together with a dyadic-tree based adaptive segmentation for

gene detection. This work considers DNA as a piecewise

stationary series, and provide a thorough mathematical foun-

dation for its analysis.

Epps et al. [33] developed an integer period DFT for bio-

logical sequence processing that has some advantages in

detecting DNA periodicities. Rushdi and Tuqan [34] ana-

lyzed other trigonometric transforms as the discrete cosine

transform (DCT), the discrete sine transform (DST) and the

discrete Hartley transform (DHT), to find periodicities in

DNA sequences. They showed also a unified multirate DSP

model based on these transforms.

Berger et al. [35] analyzed the power spectrum of the ge-

nomic sequences using the Warped DFT and the Walsh Ha-

damard Transform to improve the effectiveness in detecting

periodicities. Rodríguez-Fuentes et al. [36] introduced com-

putational improvements in using the STFT to analyze ge-

nomic sequences.

The phase of the DFT has been used as well in detecting

coding regions. Kotlar and Lavner [37] introduced the Spec-

tral Rotation Measure, deriving a method in which the DFT

phase is computed at the 1/3 frequency for the binary se-

quences for A, T, C, and G. Experimental analysis of the

genes of S. cerevisiae and other organisms showed a distri-

bution of the phase in a bell-like curve around a central value

in all four nucleotides, and a nearly uniform distribution in

the non-coding regions, allowing to define measures to iden-

tify coding regions based on this phase property. Rushdi and

Tuqan [38] derived the filtered spectral rotation measure

based on the polyphase filtered DNA spectrum introduced in

[23], as an alternative measure to detect coding regions.

Yin and Yau [39] introduced an algorithm called Exon

Prediction via Nucleotide Distributions (EPND), which

combines the information from the peak at the N/3 frequency

in the DFT and the frequencies of occurrence of the nucleo-

tides in the three codon positions (position asymmetry meas-

ure) obtaining an improvement of the effectiveness in the

detection of coding regions.

Akhtar et al. [40] showed an optimization of the period-3

methods taking into account both computational complexity

and the relative accuracy of gene prediction. In this work, a

paired and weighted spectral rotation (PWSR) measure pre-

viously defined by the authors was employed. This study

used as additional information the statistical property of eu-

karyotic sequences by which introns are rich in nucleotides

‘A’ and ‘T’ whereas exons are rich in nucleotides ‘C’ and

‘G’.

At this point, it is worth to mention that other studies like

that of Xing et al. [41] reveal that the PSD itself does not

provide sufficient resolving power to detect periodic signals

in short coding sequences, and consequently other ap-

proaches in addition to the DFT have been used for this pur-

pose.

Autoregressive modeling of DNA sequences was ad-

dressed by Chakravarthy et al. [42] who presented a model

in which AR parameters are used as features. The AR resid-

ual error analysis shows a high specificity of coding DNA

sequences, and the analysis based in AR features was useful

in distinguishing between coding and non-coding DNA se-

quences. The AR model was very specific to the coding

DNA sequences, and its specificity increased with increasing

model orders. Rao and Shepherd [43] addressed the problem

of detecting 3-periodicity in short genomic sequences based

on the AR technique, in an effort to take advantage of the

inherent improved frequency resolution of the AR models.

Akhtar et al. [44] presented an autoregressive modelling

for the classification of genomic sequences, that provides a

compact multi-dimensional feature that characterize the short

term spectrum. The AR feature was also combined with a

time-frequency hybrid (TFH) feature composed by the

PWSR measure and the time-domain average magnitude

difference function (AMDF). A Gaussian mixture model clas-

sifier was employed and showed improved recognition capa-

bilities. Another approach based on Singular Value Decom-

position was presented by the same authors in [45]. Akhtar

[46] also presents a comparison between time and frequency

domain techniques to detect short coding regions and show

some advantages of the former.

Cristea et al. [47] address the detection of nucleotide se-

quences using a two step procedure comprising a Principal

Components Analysis (PCA) stage, which retains only the

high variance components of the input signal, and a feed-

forward Artificial Neural Network (ANN), which performs

the prediction. It is shown that the PCA stage performs an

approximate DFT, passing from the time (space) domain to

the frequency domain, and the ANN implements the inverse

DFT, generating the estimate of the next sample of the se-

quence in the time (space) domain. Rodríguez-Fuentes et al.

[48] used a combination of DSP approaches to detect coding

regions in genomic sequences and showed the advantages of

the combined method over the individual ones. Gunawan et

al. [49] introduced a signal boosting technique to enhance

Page 9

36 Current Bioinformatics, 2009, Vol. 4, No. 1 Lorenzo-Ginori et al.

the coding region and improve the likelihood of its correct

identification. The authors claim that when using this

method together with ANN classification, the ratio of coding

to non-coding energy is almost doubled.

Reading frame identification is an important issue in the

detection of coding regions. This topic has also received at-

tention from the DSP point of view. Anastassiou [12] and

Kotlar and Lavner [37] presented algorithms for this pur-

pose, which make use of the phase properties of the

weighted transformed indicator sequences and showed good

results.

Detection of Coding Regions and Other Applications Us-

ing an Information Theory Approach

The concept of entropy as it is used in Information The-

ory has been employed as well to detect coding regions.

Román-Roldán et al. [50] defined a complexity measure,

based on the entropic segmentation of DNA sequences into

homogeneous domains. Bernaola-Galván et al. [51] intro-

duced a computational approach to finding borders between

coding and non-coding DNA, in which the sequences are

described by a 12-letter alphabet, capable of representing the

differential base composition at each codon position, and the

borders are searched by means of an entropic segmentation

through the Jensen-Shannon measure. The method showed to

be very accurate and does not require prior training.

Nicorici and Astola [52] extended this approach by ap-

plying recursively an entropic segmentation method on DNA

sequences using 12 and 18-symbol alphabets to capture the

differential nucleotide composition in codons as well as the

differential stop-codon in all phases of both strands. The

method uses the Jensen-Rényi divergence measure, nucleo-

tide statistics and stop codon statistics in the two DNA

strands in order to find the borders between the coding and

non-coding regions. This method does not require prior

training and showed good results.

Multihac et al. [53] used a more theoretical information

theory perspective to interpret the amount of information

carried by the binding site patterns in the DNA molecules,

using maximum entropy methods. Benson [54] defined a new

distance measure for comparing sequence profiles by esti-

mating path lengths along an entropy surface and used it to

analyze similarities within families of tandem repeats in the

C. elegans genome. May et al. [55] reviewed the existing

coding (both source and channel) theoretic methods for

modelling genetic systems, and present research results for

Escherichia coli K-12. As a last reference to be cited in this

area, Hussinia et al. [56] analyzed in a formalized mathe-

matical framework the properties of the languages used in

DNA computations.

Relative Merits of Different Approaches to Detect Cod-

ing Regions in Genomic Sequences

The methods to detect coding regions in genomic se-

quences based in finding regions with a remarkable period-3

component in the frequency spectrum, constitute a qualita-

tively different approach that is independent from other

methods (for example statistical) applied so far to solve this

task. Among the methods based in spectral analysis, the

DFT-based Spectral Rotation Measure, the Paired and

Weighted Spectral Rotation (PSWR) measure, as well as the

paired spectral content (PSC) outperforms the conventional

1-D frequency-domain methods (i. e. the simple detection of

the period-3 spectral component in its various forms), pro-

ducing higher values of specificity. By comparison with

other period-3 based measures, [15] reports that the DFT-

based PWSR measure method showed significant improve-

ments, respectively, over the Spectral Content and Spectral

Rotation measures in the detection of exonic nucleotides at a

fixed false positive rate.

Other classical methods based in the period-3 detection

like the antinotch filter and the autoregressive (AR) models

showed lower coding region detection capabilities. Formal

evaluations made in [15] revealed that the more recent

AMDF time domain method performs better in terms of ex-

onic nucleotide detection rates than the classical period-3

methods. The limitations of the classical methods in this case

have been attributed to their relatively large window size,

which reduces the time resolution. It has been suggested that

the optimum window length for period-3 based methods de-

pends on the length of the exon regions and that further im-

provements over the previously discussed methods are ob-

tained using the time-frequency hybrid method (TFH). The

authors consider that a promising line of development is the

use of combined methods in which the detection capabilities

of the combination outperforms that of the individual meth-

ods included, an approach that was used in [48].

Other Studies on Genomic Sequences Using DSP Tech-

niques

There are other characteristics of the genomic sequences

that have been studied using DSP techniques. One example

is the general analysis of latent periodicities in genomic se-

quences which appears in Arora et al. [57], where sequential

averaging is used when the data exhibits cyclostationarity

properties.

Cristea [58, 59] studied the behaviour of the phase for

complex representations of the bases in genomic sequences.

These papers report the existence of a global helicoidal

wrapping of the complex representations of the bases along

the sequences. This is considered as a large scale trend of

genomic signals. Here other properties are analyzed as well,

related to the cumulated and unwrapped phase. These theo-

retical concepts were applied by Cristea et al. [60] to identify

HIV Protease (PR) and Reverse Transcriptase (RT) muta-

tions leading to multiple drug resistance to PR and RT in-

hibitors.

Bouaynaya and Schonfeld [61, 62] analyze the long-

range power-law correlations detected in eukaryotic DNA,

introducing new non-stationary methods to study the correla-

tion properties in genomic sequences. They defined a quanti-

tative measure of the degree of randomness (deviation from

a white Gaussian process) derived from the Hilbert transform

spectrum. It was shown there that DNA sequences exhibit

long range correlations and that DNA correlations are much

more complex than power laws with a single scaling expo-

nent.

Page 10

Digital Signal Processing in the Analysis of Genomic Sequences Current Bioinformatics, 2009, Vol. 4, No. 1 37

The Discrete Wavelet Transform has been used to ana-

lyze genomic sequences. A general perspective on the use of

Wavelets and the DWT in Bioinformatics is presented by

Liò [63]. An introductory analysis of genomic sequences

using the DWT was presented by Ning et al. [64], and an

approach to visualize regular patterns in DNA was intro-

duced by Dodin et al. [65].

Referring to other various applications, Buchner and Jan-

jarasjitt [66] introduced an algorithm based on processing a

DNA sequence with the short-time periodicity transform, to

detect and visualize tandem repeats in DNA sequences, Cris-

tea et al. [67] use DSP methods for trend extraction from sets

of genomic signals and apply their methodology to study the

mutations in pathogen genomes, and Akhtar [15] evaluated

different DSP methods to detect splice sites.

Sharma et al. [68] studied the repetitive DNA sequences

using the DFT to identify significant periodicities present

and providing a complete detection of repeats together with

interactive and detailed visualization of the spectral analysis.

Dasgupta et al. [69] combined wavelet transform and

Hidden Markov Models to identify the location of CpG is-

lands in Human Genome. Another DSP approach for the

same purpose was introduced by Rushdi and Tuqan [70].

Gupta et al. [71] devised an efficient algorithm to detect pal-

indromes in DNA sequences using a signal processing opera-

tion called periodicity transform. Providence [72] applied

time-varying cellular automata to the problem of finding

signals in DNA sequences. Zhang and Kinsner [73] em-

ployed a multifractal analysis to DNA feature extraction,

using the Rényi and Mandelbrot fractal dimension spectra

for extracting the information contained in the DNA se-

quences.

Su et al. [74] applied the matched filter algorithm to ana-

lyze the structure of genomic sequences, in particular to lo-

cate and align similar segments between two sequences. An-

drade and Manolakos [75] addressed the application of DSP to

the electrophoresis process used in DNA sequencing and de-

veloped algorithms for signal background estimation and

baseline correction.

Other DSP applications related to studies on proteins can

be found in Hong and Tewfik [76], Aydin and Altunbasak

[77], Lazovic [78], Ramachandran and Antoniou [79] and

D’Avenio et al. [80].

New Perspectives of DSP Applications Based on the Al-

gebraic Structures of the Genetic Code

The numerical representation of the genetic code and

consequently of genomic sequences as has been presented in

the various references cited in this article are not unique and

extraordinary. In fact, the genetic codification systems that

have been used so far, could be non-optimum. The nature of

the genetic code is now fairly well known and there are

trends to improve predictions. From the second half of 20th

century, many attempts have been made to understand the

internal regularity of the genetic code, based on several

mathematical or geometrical points of view, by Bashford and

Jarvis [81], Bashford et al. [82], Beland and Allen [83],

Crick [84], Eck [85], Epstein [86], Jimenez-Montaño [87],

Jukes [88] and Hornos and Hornos [89]. In any case the

Code represents an extension of the four-letter alphabet of

deoxyribonucleic (DNA) bases: A, G, C, T (U in RNA).

In recent years, the genetic code algebraic structures have

been introduced by Sánchez et al. [90-92]. It has been shown

that this code constitutes a more fundamental concept than a

“conventional codification system”, as a consequence of its

biological meaning. Depending on the algebraic operation

defined in the base set, different structures were obtained. If

the Watson-Crick base pairing (G:C and A:T) is expressed

by the classical logical operations with “OR” (?) and “AND”

(?) in such a way that the following expressions hold:

G?C=C, T?A=C, G?C=G and T?A=G then a Boolean alge-

bra is obtained which is isomorphic to the Boolean algebra

defined on the set {0,1}2: G?00, A?01, T?10 and C?11

[90]. This leads to a binary representation of DNA se-

quences. On the other hand, if the Watson-Crick base pairing

is expressed by the sum “+”: G+C=C and U+A=C then this

requirement leads to define an additive group on the DNA

base set, isomorphic to the complex representation: G?1,

A?exp(?i/2), T?exp(?i) and C?exp(3?i/2) [92].

Notice that here a numerical representation of DNA

bases refer to their algebraic representation, which means the

existence of an isomorphism between an algebraic structure

with a biological meaning defined in the base or codon sets,

and another one defined in some numerical set. We point out

that the numerical representations mentioned before in this

paper are codification (ad hoc) but not algebraic representa-

tions because of the absence of algebraic operations. These

new models lead to go beyond the genetic code limits to deal

with the quantitative relationship between DNA genomic

sequences.

In particular, the extension of the four DNA base set with

a dummy variable (D) leads to analogous algebraic struc-

tures, useful to deal with the multiple sequence alignments of

genomic regions where the gaps are replaced by the symbol

D [93]. For instance, the additive group defined in the set

{D, G, A, T, C} is isomorphic to the complex representation:

D ? 1, G ? exp(2?i/5), A ? exp(4?i/5), T ? exp(6?i/5)

and C ? exp(8?i/5). The 3-periodicity was detected this way

in the power spectra of the complex representations of mul-

tiple aligned genomes from HIV-1 [94]. These results

showed the theoretical possibilities of using generalized DSP

techniques in the comparative genomics.

CONCLUSION

The application of Digital Signal Processing in Genomic

Sequence Analysis has received great attention in the last

few years, providing a new insight in the solution of various

problems like

•

Detection of coding regions in genomic sequences

based on spectral analysis.

•

Reading frame identification.

•

Detection of periodicities in genomic sequences.

•

Detection of CpG islands.

•

Detection of palindromes.

Page 11

38 Current Bioinformatics, 2009, Vol. 4, No. 1 Lorenzo-Ginori et al.

•

Finding diverse signals and features in genomic se-

quences.

•

Studies on proteins.

On the other hand, the main DSP tools that have found

application in this field are

•

Digital filters (IIR, FIR).

•

Discrete transforms (Fourier, Cosine, Walsh Ha-

damard, Wavelet).

•

Parametric models (mainly autoregressive).

•

Information Theory concepts (entropy).

•

Fractals.

Other algorithmic tools that have been applied in Bioin-

formatics although not addressed in this paper are considered

usually as neighbouring areas. This is the case of Hidden

Markov Models (HMM), Artificial Neural Networks (ANN),

Support Vector Machines (SVM), Fuzzy Sets and Genetic

Algorithms.

A recent development closely related to the impact of

DSP on Bioinformatics is the new field of Genomic Signal

Processing (GSP). An early survey on this can be found in

Zhang et al. [95]. A formal definition of GSP was given by

Dougherty et al. [96] as “the analysis, processing, and use of

genomic signals for gaining biological knowledge and the

translation of that knowledge into systems-based applica-

tions.” Schonfeld et al. [97] remark the current interest in

using DSP methods to obtain information from genomic and

proteomic data to build models of molecular biological sys-

tems. This would allow obtaining a deeper understanding of

the structure and functions of living systems and will help in

developing new diagnostic tools, therapeutic procedures and

pharmacological drugs. An application example in cancer

classification and prediction can be seen in Qiu et al. [98].

Finally, it is interesting to notice that Bioinformatics is

also having an influence on new developments, as can be

seen in [99, 100].

ACKNOWLEDGEMENTS

The authors wish to acknowledge the constructive com-

ments and critical reading of the manuscript made by the

anonymous reviewers.

This research was partially funded by the Canadian In-

ternational Development Agency Project Tier II-394-TT02-

00 and by the Flemish VLIR-UOS Programme for Institu-

tional University Co-operation (IUC).

REFERENCES

[1]

De Fonzo V, Aluffi-Pentini F, Parisi V. Hidden Markov Models in

Bioinformatics. Curr Bioinform 2007; 2: 49-61.

Oppenheim AV, Schafer R. Discrete-Time Signal Processing (3rd

Edition), Prentice-Hall, NY 2009.

Proakis JG, Manolakis DK, Digital Signal Processing (4th Edi-

tion), Prentice Hall, NY 2006.

Stoica P, Moses RL. Spectral Analysis of Signals, Prentice-Hall,

NY 2005.

Burrus CS, Gopinath RA, Guo H. Introduction to Wavelets and

Wavelet Transforms: A Primer, Prentice-Hall, NY 1997.

[2]

[3]

[4]

[5]

[6]

Shannon CE. A Mathematical Theory of Communication. he Bell

Sys Techn J 1948; 27: 379-23, 623-56.

Rényi A. On measures of information and entropy. Proceedings of

the 4th Berkeley Symposium on Mathematics, Statistics and

Probability 1960: 547-61.

Voss RF. Evolution of long-range fractal correlations and 1/f noise

in DNA base sequences. Phy. Rev. Lett 1992; 68: 3805-08.

Zhang R, Zhang CT. Z curves, an intuitive tool for visualizing and

analyzing the DNA sequences. J. Biomol. Struct. Dyn. 1994; 11:

767-82.

Rushdi A, Tuqan J. Gene Identification Using the Z-Curve

Representation. Proceedings of the IEEE International Conference

on Acoustics, Speech and Signal Processing, ICASSP 2006; 2: II-

II. Cristea PD . Genomic signal analysis: Study of pathogen variabil-

ity. Proceedings of the IEEE International Workshop on Genomic

Signal Processing and Statistics GENSIPS '06 2006: 51-52.

Anastassiou D. Genomic signal processing. IEEE Sign Proc Mag

2001; 18: 8-20.

Cristea P, Deklerck R, Cornelis J, Tuduce R, Nastac I, Andrei M.

Signal Representation and Processing of Nucleotide Sequences.

Proceedings of the 7th IEEE International Conference on Bioin-

formatics and Bioengineering, BIBE 2007: 1214-19.

Cristea PD, Tuduce R, Cornelis J. Signal Analysis of Pathogens

Genomic Sequences. Frontiers in the Convergence of Bioscience

and Information Technologies, FBIT 2007: 245- 50.

Akhtar M, Epps J, Ambikairajah E. Signal Processing in Sequence

Analysis: Advances in Eukaryotic Gene Prediction. IEEE J Select

Topics Sign Proc 2008; 3: 310-21.

Cristea PD. Genetic Signal Analysis. Proc Int Symp Sign Proc Appl

(ISSPA) 2001; 2: 703-08.

Rushdi A, Tuqan J. The role of the symbolic-to-numerical mapping

in the detection of DNA periodicities. Proceedings of the IEEE In-

ternational Workshop on Genomic Signal Processing and Statis-

tics, GENSIPS '08 2008: 1-4.

Brodzik AK, Peters O. Symbol-balanced quaternionic periodicity

transform for latent pattern detection in DNA sequences. Proceed-

ings IEEE International Conference onAcoustics, Speech, and Sig-

nal Processing, (ICASSP '05) 2005; 5: 373-76.

Trifonov EN. 3-, 10.5-, 200- and 400-base periodicities in genome

sequences. Physica A 1998; 249: 511-16.

Tuqan J, Rushdi A. A DSP Approach for Finding the Codon Bias

in DNA Sequences. IEEE J Select Topics Sign Proc 2008; 2: 343-

56.

Sánchez J, López-Villaseñor I. A simple model to explain three-

base periodicity in coding DNA. FEBS Lett 2006; 580: 6413-22.

Vaidyanathan PP, Yoon BJ. The role of signal-processing concepts

in genomics and proteomics. J Franklin Inst 2004; 341: 111-35.

Tuqan J, Rushdi A. A DSP perspective to the period-3 detection

problem. Proceedings of the IEEE International Workshop on Ge-

nomic Signal Processing and Statistics, GENSIPS '06 2006: 53-54.

Fox TW, Carreira A. A Digital Signal Processing Method for Gene

Prediction with Improved Noise Suppression. EURASIP J Appl

Sign Proc 2004; 1: 108-11.

Afreixo V, Ferreira PJSG, Santos D. Fourier analysis of symbolic

data: A brief review. Digit Sign Proc 2004; 14: 523-30.

Tiwari S, Ramachandran S, Bhattacharya A, Bhattacharya S,

Ramaswamy R. Prediction of probable genes by Fourier analysis of

genomic sequences. CABIOS 1997; 13: 263-70.

Yan M, Lin ZS, Zhang CT. A new Fourier Transform approach for

protein coding measure based on the format of the Z curve.

Bioinformatics 1998; 14: 685-90.

Datta S, Asif A, Wang H. Prediction of protein coding regions in

DNA sequences using Fourier spectral characteristics. Proceedings

of the IEEE Sixth International Symposium on Multimedia Software

2004: 160-63.

Datta S, Asif A. A fast DFT based gene prediction algorithm for

identification of protein coding regions. Proceedings IEEE Interna-

tional Conference on Acoustics, Speech, and Signal Processing,

(ICASSP '05) 2005; 5: 653-56.

Datta S, Asif A. DFT based DNA splicing algorithms for prediction

of protein coding regions. Proceedings of the IEEE Thirty-Eighth

Asilomar Conference on Signals, Systems and Computers 2004; 1:

45-49.

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

Page 12

Digital Signal Processing in the Analysis of Genomic Sequences Current Bioinformatics, 2009, Vol. 4, No. 1 39

[31]

Isaac B, Singh H, Kaur H, Raghava GPS. Locating probable genes

using Fourier Transform approach. Bioinform Appl Note 2002; 18:

196-97.

Stoffer D, Ombao HC, Tyler DE. Local spectral envelope: an ap-

proach using dyadic tree-based adaptive segmentation. Ann Inst

Statist Math 2002. 54: 201-23.

Epps J, Ambikairajah E, Akhtar M. An integer period DFT for

biological sequence processing. Proceedings of the IEEE Interna-

tional Workshop on Genomic Signal Processing and Statistics

GENSIPS 2008: 1-4.

Rushdi A, Tuqan J. Trigonometric transforms for finding repeats in

DNA sequences. Proceedings of the IEEE International Workshop

on Genomic Signal Processing and Statistics, GENSIPS '08 2008:

1-4.

Berger JA, Mitra SK, Astola J. Power spectrum analysis for DNA

sequences. Proc IEEE Seventh Int Symp Sign Proc Appl 2003; 2:

29-32.

Rodríguez-Fuentes A, Lorenzo-Ginori JV, Grau-Ábalo R. Detec-

tion of coding regions in large DNA sequences using the short time

Fourier Transform. Lect Notes Comput Sci 2006. 4225: 902-909.

Kotlar D, Lavner Y. Gene Prediction by Spectral Rotation Meas-

ure: A New Method for identifying Protein-Coding Regions. Ge-

nome Res 2003; 13: 1930-1937.

Rushdi A, Tuqan J. The Filtered Spectral Rotation Measure. Proc

the IEEE Fortieth Asilomar Conf Sign Sys Comput ACSSC '06

2006: 1875-79.

Yin C, Yau SS-T. Prediction of protein coding regions by the 3-

base periodicity analysis of a DNA sequence. J Theor Biol 2007;

247: 687-94.

Akhtar M, Ambikairajah E, Epps J. Optimizing period-3 methods

for eukaryotic gene prediction. Proceedings of the IEEE Interna-

tional Conference on Acoustics, Speech and Signal Processing IC-

ASSP 2008: 621-24.

Xing C, Bitzer DL, Alexander WE, Stomp AM, Vouk MA. Free

Energy Analysis on the Coding Region of the Individual Genes of

Saccharomyces cerevisiae. Proceedings of the 28th IEEE EMBS

Annual International Conference 2006: 4225-28.

Chakravarthy N, Spanias A, IasemidisLD, Tsakalis K. Autoregres-

sive Modeling and Feature Analysis of DNA Sequences. EURASIP

J Appl Sign Proc 2004; 1: 13-28.

Rao N, Shepherd SJ. Detection of 3-periodicity for small genomic

sequences based on AR technique. Proceedings of the IEEE

International Conference on Communications, Circuits and

Systems, ICCCAS 2004; 2: 1032-36.

Akhtar M, Ambikairajah E, Epps J. Comprehensive autoregressive

modeling for classification of genomic sequences. Proceedings of

the IEEE 6th International Conference on Information,

Communications & Signal Processing 2007: 1-5.

Akhtar M, Ambikairajah E, Epps J. Detection of Period-3 Behavior

in Genomic Sequences Using Singular Value Decomposition. Pro-

ceedings of the IEEE International Conference on Emerging Tech-

nologies 2005: 13-17.

Akhtar M. Comparison of Gene and Exon Prediction Techniques

for Detection of Short Coding Regions. Int J Inform Technol 2005;

8: 26-35.

Cristea P, Deklerck R, Cornelis J, Tuduce R, Nastac I, Andrei M.

ANN Prediction of Nucleotide Sequences Link of Principal Com-

ponent Analysis to Fourier Transform. Proceedings of the 14th In-

ternational Workshop on Systems, Signals and Image Processing,

and 6th EURASIP Conference focused on Speech and Image Proc-

essing, Multimedia Communications and Services 2007: 69-73.

Rodríguez-Fuentes A, Lorenzo-Ginori JV, Grau-Ábalo R. Coding

Region Prediction in Genomic Sequences Using a Combination of

Digital Signal Processing Approaches. Lect Notes Comput Sci

2007; 4756: 635-642.

Gunawan TS, Ambikairajah E, Epps J. Boosting approach to exon

detection in DNA sequences. Electron Lett 2008; 44: 323-24.

Román-Roldán R, Bernaola-Galván P, Oliver JL. Sequence

Compositional Complexity of DNA through an Entropic

Segmentation Method. Phys Rev Lett 1998. 80: 1344-47.

Bernaola-Galván P, Grosse I, Carpena P, Oliver JL, Román-Roldán

R, Stanley HE. Finding Borders between Coding and Noncoding

DNA Regions by an Entropic Segmentation Method. Phys Rev Lett

2000; 85: 1342-45.

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

Nicorici D, Astola J. Segmentation of DNA into Coding and Non-

coding Regions Based on Recursive Entropic Segmentation and

Stop-Codon Statistics. EURASIP J Appl Sign Proc 2004; 1: 81-91.

Mutihac R, Cicuttin A, Mutihac RC. Entropic approach to informa-

tion coding in DNA molecules. Mat Sci Eng C 2001; 18: 51-60.

Benson G. A new distance measure for comparing sequence pro-

files based on path lengths along an entropy surface. Bioinformatics

2002; 18 (Suppl 2): S44-S53.

May EE, Vouk MA, Bitzer DL, Rosnick DI. An error-correcting

code framework for genetic sequence analysis. J Franklin Inst

2004; 341: 89-109.

Hussinia S, Karib L, Konstantinidisa S. Coding properties of DNA

languages. Theor Comput Sci 2003; 290: 1557-79.

Arora R, Sethares WA, Bucklew JA. Latent Periodicities in Ge-

nome Sequences. IEEE J Select Topics Sign Proc 2008; 3: 332-42.

Cristea PD. Phase analysis of DNA genomic signals. Proceedings

of the 2003 Int Symp Circuits Sys ISCAS '03 2003; 5: 25-28.

Cristea PD. Multiresolution phase analysis of genomic signals.

Proceedings of the First International Symposium on Control,

Communications and Signal Processing 2004: 743-46.

Cristea PD, Tuduce RA. Otelea D. Study of HIV Variability based

on Genomic Signal Analysis of Protease and Reverse Transcriptase

Genes. Proceedings of the 27th Annual International Conference of

the IEEE Engineering in Medicine and Biology Society 2005:

4795-98.

Bouaynaya N, Schonfeld D. Emergence of new structure from non-

stationary analysis of genomic sequences. Proceedings of the IEEE

International Workshop on Genomic Signal Processing and Statis-

tics GENSiPS 2008: 1-4.

Bouaynaya N, Schonfeld D. Nonstationary Analysis of Coding and

Noncoding Regions in Nucleotide Sequences. IEEE J Select Topics

in Sign Proc 2008; 3: 357-64.

Liò P. Wavelets in bioinformatics and computational biology: state

of art and perspectives. Bioinform Rev 2003; 19: 2-9.

Ning J, Moore CN, Nelson JC. Preliminary wavelet analysis of

genomicsequences. Proceedings of the IEEE Bioinformatics Con-

ference 2003; 509-10.

Dodin G, Vandergheynst P, Levoir P, Cordier C, Marcourt L. Fou-

rier and Wavelet Transform Analysis, a Tool for Visualizing Regu-

lar Patterns in DNA Sequences. J Theor Biol 2000; 206: 323-26.

Buchner M, Janjarasjitt S. Detection and Visualization of Tandem

Repeats in DNA Sequences. IEEE Transac Sign Proc 2003; 51:

2280-87.

Cristea P, Tuduce R, Monteanu A, Cornelis J. Common Trend

Extraction from Sets of Genomic Signals. Proc IEEE Int Symp

Commun Control Sign Proc ISCCSP 2008: 1205-10.

Sharma D, Issac B, Raghava GPS, Ramaswamy R. Spectral Repeat

Finder (SRF): identification of repetitive sequences using Fourier

transformation. Bioinformatics 2005; 20: 1405-12.

Dasgupta N, Lin S, Carin L. Sequential Modeling for Identifying

CpG Island Locations in Human Genome. IEEE Sign Proc Lett

2002; 9: 407-09.

Rushdi A, Tuqan J. A New DSP-Based Measure for CPG Islands

Detection. Proceedings of the IEEE 12th Signal Processing Educa-

tion Workshop, 4th Digital Signal Processing Workshop 2006: 561-

65.

Gupta R, Mittal A, Gupta S. An efficient algorithm to detect palin-

dromes in DNA sequences using periodicity transform. Signal

Processing 2006; 86: 2067-73.

Providence SV. Utilization of Cellular Automata in the DNA Sig-

nal Search Problem. Proc IEEE Southeast Conf 2004; 325-29.

Zhang H, Kinsner W. Feature extraction from DNA sequences by

multifractal analysis. Proceedings of the 23rd Annual International

Conference of the IEEE Engineering in Medicine and Biology So-

ciety 2001; 1567-72.

Su SC, Yeh CH, Kuo CCJ. Structural analysis of genomic se-

quences with matched filtering. Proceedings of the 25th Annual In-

ternational Conference of the IEEE Engineering in Medicine and

Biology Society 2003; 3: 2893-96.

Andrade L, Manolakos ES. Signal Background Estimation and

Baseline Correction Algorithms for Accurate DNA Sequencing. J

VLSI Sign Proc 2003; 35: 229-43.

Hong C, Tewfik AH. Efficient Updating of Biological Sequence

Analyses. IEEE J Select Topics Sign Proc 2008; 2: 365-77.

[53]

[54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

[70]

[71]

[72]

[73]

[74]

[75]

[76]

Page 13

40 Current Bioinformatics, 2009, Vol. 4, No. 1 Lorenzo-Ginori et al.

[77]

Aydin Z, Altunbasak Y. A signal processing application in ge-

nomic research: protein secondary structure prediction. IEEE Sign

Proc Mag 2006; 23: 128-31.

Lazovic J. Selection of amino acid parameters for Fourier trans-

form-based analysis of proteins. CABIOS COMMUNICATION

1996; 12: 553-62.

Ramachandran P, Antoniou A. Identification of Hot-Spot Locations

in Proteins Using Digital Filters. IEEE J Select Topics Sign Proc

2008; 2: 378-89.

D’Avenio G, Grigioni M, Orefici G, Creti R. SWIFT (sequence-

wide investigation with Fourier transform): a software tool for

identifying proteins of a given class from the unannotated genome

sequence. Bioinform 2005; 21: 2943-49.

Bashford JD, Jarvis PD. The genetic code as a periodic table. Bio-

systems 2000; 57: 147-161.

Bashford JD, Tsohantjis I, Jarvis PD. A supersymmetric model for

the evolution of the genetic code. Proc Natl Acad Sci USA 1998;

95: 987-992.

Beland P, Allen TF. The origin and evolution of the genetic code. J

Theor Biol 1994; 170: 359-365.

Crick FHC. The origin of the genetic code. J Mol Biol 1968; 38:

367-379.

Eck RV. Genetic code – Emergence of a symmetrical pattern. Sci-

ence 1963; 140: 477-481.

Epstein CJ. Role of the amino-acid “code” and of selection for

conformation in the evolution of proteins. Nature 1966; 210: 25-28.

Jimenez-Montaño MA. The hypercube structure of the genetic code

explains conservative and non-conservative amino acid substitu-

tions in vivo and in vitro. Biosystems 1996; 39: 117-125.

Jukes TH. The amino acid code. In A. Neuberger, Comprehensive

Biochemistry. Elsevier Amsterdam 1977, pp. 235-293.

[78]

[79]

[80]

[81]

[82]

[83]

[84]

[85]

[86]

[87]

[88]

[89]

Hornos JE, Hornos YM. Algebraic model for the evolution of the

genetic code. Phys Rev Lett 1993; 71: 4401-4404.

Sánchez R, Morgado E, Grau R. A genetic code boolean structure

I. The meaning of boolean deductions. Bull Math Biol 2005; 67: 1-

14.

Sánchez R, Morgado E, Grau R. Gene algebra from a genetic code

algebraic structure. J Math Biol 2005; 51, 431-457,

Sánchez R, Grau R. A novel algebraic structure of the Genetic

Code over the Galois Fields of four DNA Bases. Acta Biotheo-

retica 2006; 54: 27-42.

Sánchez R, Grau R, Morgado E. A Novel DNA Sequence Vector

Space over an extended Genetic Code Galois Field. MATCH Com-

mun Math Comput Chem 2006; 56: 5-20.

Sanchez R., Grau R. An algebraic hypothesis about the primeval

genetic code. Second international workshop Cuba/Flanders

IWOBI 08. ISBN 978-959-250-394-6. arXiv:0805.1128v3

Zhang XY, Chen F, Zhang YT, et al. Signal Processing Techniques

in Genomic Engineering. Proc IEEE 2002; 90: 1822-33.

Dougherty ER, Datta A, Sima C. Research issues in genomic signal

processing. IEEE Sign Proc Mag 2005; 22: 46-48.

Schonfeld D, Goutsias J, Shmulevich I, Tabus I, Tewfik AH. Intro-

duction to the Issue on Genomic and Proteomic Signal Processing.

IEEE J Select Topics Signl Proc 2008; 2: 257-59.

Qiu P, Wang Z J, Liu KJR. Genomic Processing for Cancer Classi-

fication and Prediction. IEEE Sign Proc Mag 2007; 24: 100-10.

Chen J, Li H, Sun K, Kim B. How will bioinformatics impact sig-

nal processing research? IEEE Sign Proc Mag 2003; 6: 106-26.

Tsaftaris SA, Katsaggelos AK, Pappas TN, Papoutsakis ET. How

Can DNA Computing be Applied to Digital Signal Processing?

IEEE Sign Proc Mag 2004; 21: 57-61.

[90]

[91]

[92]

[93]

[94]

[95]

[96]

[97]

[98]

[99]

[100]

Received: September 29, 2008

Revised: October 14, 2008 Accepted: October 28, 2008