Identification of biomarkers from mass spectrometry data using a "common" peak approach.

Abstract

Proteomic data obtained from mass spectrometry have attracted great interest for the detection of early-stage cancer. However, as mass spectrometry data are high-dimensional, identification of biomarkers is a key problem.
This paper proposes the use of "common" peaks in data as biomarkers. Analysis is conducted as follows: data preprocessing, identification of biomarkers, and application of AdaBoost to construct a classification function. Informative "common" peaks are selected by AdaBoost. AsymBoost is also examined to balance false negatives and false positives. The effectiveness of the approach is demonstrated using an ovarian cancer dataset.
Continuous covariates and discrete covariates can be used in the present approach. The difference between the result for the continuous covariates and that for the discrete covariates was investigated in detail. In the example considered here, both covariates provide a good prediction, but it seems that they provide different kinds of information. We can obtain more information on the structure of the data by integrating both results.

Full text

Page 1

BioMed Central
Page 1 of 9
(page number not for citation purposes)
BMC Bioinformatics
Open Access
Methodology article
Identification of biomarkers from mass spectrometry data using a
"common" peak approach
Tadayoshi Fushiki*, Hironori Fujisawa and Shinto Eguchi
Address: Department of Mathematical Analysis and Statistical Inference, Institute of Statistical Mathematics, Tokyo, Japan
Email: Tadayoshi Fushiki* - fushiki@ism.ac.jp; Hironori Fujisawa - fujisawa@ism.ac.jp; Shinto Eguchi - eguchi@ism.ac.jp
* Corresponding author
Abstract
Background: Proteomic data obtained from mass spectrometry have attracted great interest for
the detection of early-stage cancer. However, as mass spectrometry data are high-dimensional,
identification of biomarkers is a key problem.
Results: This paper proposes the use of "common" peaks in data as biomarkers. Analysis is
conducted as follows: data preprocessing, identification of biomarkers, and application of AdaBoost
to construct a classification function. Informative "common" peaks are selected by AdaBoost.
AsymBoost is also examined to balance false negatives and false positives. The effectiveness of the
approach is demonstrated using an ovarian cancer dataset.
Conclusion: Continuous covariates and discrete covariates can be used in the present approach.
The difference between the result for the continuous covariates and that for the discrete covariates
was investigated in detail. In the example considered here, both covariates provide a good
prediction, but it seems that they provide different kinds of information. We can obtain more
information on the structure of the data by integrating both results.
Background
Mass spectrometry is being used to generate protein pro-
files from human serum, and proteomic data obtained
from mass spectrometry have attracted great interest for
the detection of early-stage cancer (for example, [1-3]).
Recent advancements in proteomics come from the devel-
opment of protein mass spectrometry. Matrix-assisted
laser desorption and ionization (MALDI) and surface
enhanced laser desorption/ionization (SELDI) mass spec-
trometry provide high-resolution measurements. Mass
spectrometry data are ideally continuous data. Some
method is required to deal with high-dimensional but
small sample-size data, similar to microarray data. An
effective methodology for identifying biomarkers in high-
dimensional data is thus an important problem.
Ovarian Dataset 8-7-02, available from the National Can-
cer Institute, was analyzed in this paper. This dataset is
raw data and consists of 91 controls and 162 ovarian can-
cer patients. The mass spectrometry data for an individual
is illustrated in Fig. 1. The horizontal axis indicates the m/
z-value and the vertical axis the intensity of ion. A charac-
teristic of the data is that a number of peaks can be
observed. Each peak represents a singly charged positive
ion originating from a protein in the sample. Peaks
present in the mass spectrometry data may be usable as
biomarkers to judge whether an individual is affected or
not.
Some methodologies have been proposed for the identifi-
cation of biomarkers from ideally continuous mass spec-
Published: 26 July 2006
BMC Bioinformatics 2006, 7:358doi:10.1186/1471-2105-7-358
Received: 12 March 2006
Accepted: 26 July 2006
This article is available from: http://www.biomedcentral.com/1471-2105/7/358
© 2006 Fushiki et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 2

BMC Bioinformatics 2006, 7:358http://www.biomedcentral.com/1471-2105/7/358
Page 2 of 9
(page number not for citation purposes)
trometry data. One approach is to use peaks in the data to
identify biomarkers. Yasui et al.[4] and Tibshirani et al.[5]
adopted this approach. Another approach is based on bin-
ning of data. Yu et al.[6] and Geurts et al.[7] analyzed
mass spectrometry data after binning the data. The meth-
odology presented in this paper adopts the former
approach, since peaks in mass spectrometry data are con-
sidered to represent biological information. Our idea is
that "common" peaks within the sample might contain
useful information. This means that a peak seen for only
one subject may be noise, whereas a peak exhibited by
many subjects might be useful. In this study, an ovarian
cancer dataset was analyzed as follows: (1) preprocessing,
(2) peak detection, (3) identification of biomarkers, (4)
classification. AdaBoost was used to construct a classifica-
tion function. AsymBoost was also examined for balanc-
ing the false negatives and the false positives. The
effectiveness of the approach is evaluated using validation
data.
The proposed approach is closely related to that of Yasui
et al.[4]. In [4], biomarkers were specified through classi-
fication by AdaBoost. The present approach differs in that
"common" peaks are extracted before classification to
specify biomarkers. By specifying biomarkers before clas-
sification, the dimension of covariates becomes smaller in
classification. We think that it is better if the number of
covariates is small in classification. Whereas Yasui et al.[4]
used discrete covariates, both continuous covariates and
discrete covariates can be used in the present approach,
according to the situations. Furthermore, we recommend
that the results obtained using discrete covariates should
be compared with those obtained using continuous cov-
ariates. In the example considered here, more information
on the structure of the data can be obtained through the
use of both covariates, i.e., different kinds of informative
features can be obtained by the use of both covariates.
Results and discussion
Dataset
The range of m/z-value in the dataset is approximately [0,
20000]. However, the frequency of the peaks is too high
in the interval [0,1500] and in some cases it is difficult to
derive information from peaks in the interval [0,1500].
Therefore, the dataset is analyzed only in the interval
[1500, 20000], as in [4].
As is often the case in statistical learning [8], the dataset is
divided into two sets; a training dataset that consists of 73
controls and 130 ovarian cancer patients and a test dataset
that consists of 18 controls and 32 ovarian cancer
patients. The number of the training and test datasets are
denoted by N and n, respectively (N = 203 and n = 50).
The method proposed in this paper is trained using the
training dataset, and the performance of the trained
scheme is checked using the test dataset.
Preprocessing
Proteomic data obtained from mass spectrometry are
often inaccurate in some senses. For example, the mass/
charge axis shift is a big problem in many cases [9,10].
Therefore preprocessing of the data is very important. Pre-
processing methods have recently been proposed by
Wong et al.[9] and Jeffries [10].
In this paper, preprocessing of the dataset is performed
using SpecAlign [9] as follows: (i) subtract baseline, (ii)
generate spectrum average, (iii) spectra alignment (peak
matching method). It should be noted that it is difficult to
align spectra perfectly even if some alignment algorithm is
used. In the section of Identification of biomarkers, this
problem is reconsidered.
Peak detection
The peak detection rule of Yasui et al.[4] is adopted here.
An m/z point is regarded as a peak if it takes the maximum
value in the k-nearest neighborhood. If k is small, a point
is easily recognized as a peak. An appropriate k can be
selected by examining some ks as done in Yasui et al.[4].
We empirically set k = 10. In this study, a slightly small k
is used, since only the "common" peak is considered as a
biomarker.
Identification of biomarkers
Suppose that some individuals have a peak at a certain m/
z-value, m*. It is then expected that there exists a protein
related with the ion corresponding to m*. Therefore, m*
may be a biomarker that can be used to judge whether an
individual is affected or not. But a peak exhibited by only
one subject may just be noise. The peaks "commonly"
Typical example of proteomic data
Figure 1
Typical example of proteomic data.
0
10
20
30
40
50
60
70
80
2000 4000 6000 8000 10000
m/z
12000 14000 16000 18000 20000
Intensity

Page 3

BMC Bioinformatics 2006, 7:358http://www.biomedcentral.com/1471-2105/7/358
Page 3 of 9
(page number not for citation purposes)
exhibited by many subjects are thus candidates of biomar-
kers. However, there remains the problem that the m/z-
values are not perfectly aligned in general, so that the
above idea cannot be applied directly. By overcoming this
problem of imperfect alignment, the method for identify-
ing such "common peaks" is derived in the following.
First, an "average of peaks" is constructed by averaging
Gaussian kernels with centers at peaks, as illustrated in
Fig. 2. The "average of peaks" is then expressed as
where pi,j is the m/z-value of the i-th observation and the j-
th peak and σ (pi,j) shows the "width" of the peak. In this
study, σ (pi,j) = 0.001 × pi,j. In general, a very small σ is not
desirable because the same peaks could not be "added"
properly, but a very large σ is not also desirable because a
peak affects other peaks. The value of σ could be decided
based on the accuracy of the mass/charge axis. In this
study, 2σ (pi,j)/pi,j = 0.002. This corresponds to about ±
0.2% error of the mass/charge value of each peak point. In
this dataset, a small change of σ did not affect the result.
Secondly, a biomarker is identified by finding the peak
greater than hth in the "average of peaks," where hth is a
parameter controlling how "common" peaks can be
regarded as biomarkers (Fig. 2 (e)). With hth = 0.1, 146
biomarkers were obtained by the procedure described in
the previous section (Fig. 3 (c)). The features of the
approach are as follows:
• In general, peaks cannot be aligned perfectly even if
some alignment algorithm is applied, as stated in the sec-
tion of Preprocessing. In the "average of peaks," even if
peaks are not aligned perfectly, they can be added because
they have "width" σ (pi,j) (Figs. 2 (d) and 2 (e)).
• Another possible approach is to use the average of inten-
sities (Fig. 2 (b)). However, we think that the "average of
peaks" is more effective in mass spectrometry data (Fig. 2
(e)). "Common" peaks with small intensities can be
found easily in Fig. 3 (b), whereas it is difficult to find
such "small common" peaks in Fig. 3 (a). Furthermore,
the difference between controls and ovarian cancer
patients can be seen more clearly in the "average of peaks"
than in the average of intensities (Fig. 4).
There are many ways to reduce the number of biomarkers
determined by the above procedure. One way is to select
biomarkers that are effective in classification. In this
study, however, we simply used the biomarkers obtained
by the above procedure.
A x
( )
N
xp
p
ji
N
∑
=
i j
,
i j
,
exp
()
{ (
σ
)}
,
=−
−
⎡
⎢
⎢
⎣
⎤
⎥
⎥
⎦
( )
1
∑
1
1
2
2
(a) Intensities
Figure 2
(a) Intensities. (b) Average of intensities. (c) Peaks. (d) Gaus-
sian kernels with centers at peaks. (e) "Average of peaks"
with hth = 0.3.

Page 4

BMC Bioinformatics 2006, 7:358http://www.biomedcentral.com/1471-2105/7/358
Page 4 of 9
(page number not for citation purposes)
The covariates are extracted from the data as discrete vari-
ables and/or continuous variables. Let m1, m2, ? be the m/
z-values of the biomarkers. The discrete covariate is
obtained by searching for a peak within a window of the
biomarker, i.e.,
The continuous covariate is the maximum value of the
intensity within the window.
xj = the maximum value of the intensity within [(1 - ρ)mj,
(1 + ρ)mj].
x
m
)
m
)
j
jj
=
−+
111,[(,(]
, ,
there exists a peak within a window
ρρ
,.0otherwise
⎧
⎨⎪
⎩ ⎪
(a) Average of intesities
Figure 3
(a) Average of intesities. Intensities are averaged after nor-
malization to [0, 1]. (b) "Average of peaks." (c) Selected
biomarkers with hth = 0.1. In (c), crosses denote biomarkers.
0
3000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
3200 3400 3600 3800 4000 4200 4400
average of intensities
m/z
0
3000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
3200 3400 3600 3800 4000 4200 4400
average of peaks
m/z
0
3000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
3200 3400 3600 3800 4000 4200 4400
average of peaks
h_th
x
x
x
x
x
x
x
x
x
x
x
x
xx
x
x
xx
x
x
x
x
x
m/z
(a) Average of intensities for control and ovarian cancer patient datasets
Figure 4
(a) Average of intensities for control and ovarian cancer
patient datasets. (b) "Average of peaks" for control and ovar-
ian cancer patient datasets.
0
3000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
3200 3400 3600 3800 4000 4200 4400
average of intensities (Cancer)
average of intensities (Control)
m/z
0
3000
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
3200 3400 3600 3800 4000 4200 4400
average of peaks (Cancer)
average of peaks (Control)
m/z

Page 5

BMC Bioinformatics 2006, 7:358http://www.biomedcentral.com/1471-2105/7/358
Page 5 of 9
(page number not for citation purposes)
In this study, ρ = 0.002.
AdaBoost
The present objective is to find the important features of a
peak pattern associated with a disease on the basis of peak
identification on proteomic data. We introduce AdaBoost
for the extraction of informative patterns in the feature
space based on examples consisting of N pairs of the fea-
ture vector and the label. In this context, the feature vector
is obtained from peak intensities over the detected m/z-
values for a subject, and the label expresses the disease sta-
tus of the subject. For pattern classification, one of two
cases are employed, that is, in which the feature vector is
composed of discrete or continuous values, as discussed
in the preceding section.
Ensemble learning has been studied in machine learning.
AdaBoost [11] is one of the most efficient learning meth-
ods in ensemble learning. As explained below, AdaBoost
provides a classification function by a linear combination
of weak learners. The AdaBoost algorithm can be regarded
as a sequential minimization algorithm for the exponen-
tial loss function.
Let x be a feature vector and y a binary label with values +1
and -1. A classification function f(x) is then used to predict
the label y. If f(x) is positive (or negative), then the label
is predicted as +1 (or -1). Suppose that a class of classifi-
cation functions = {f} is provided. In AdaBoost, a clas-
sification function f in is called a weak learner. A new
classification function F is constructed by taking a linear
combination of classification functions in , i.e.,
where β = (β1, ?, βT) is a weight vector and ft ∈
1, ?, T. The sign of F(x; β) provides a label prediction of
y. This is a rule of majority vote by T classification func-
tions f1(x), ?, fT(x) with weights β1, ?, βT. Consider a
problem in which weights β1, ?, βT and classification
functions f1(x), ?,fT(x) are optimally combined based on
N given examples of (x1, y1), ?, (xN, yN). AdaBoost aims to
solve the problem by minimizing the exponential loss
defined by
for t =
AdaBoost does not jointly provide the optimal solution,
but offers a sophisticated learning algorithm with sequen-
tial structure involving two stages of optimization in
which the best weak learner ft(x) is selected in the first
stage and the best scalar weight βt is determined in the sec-
ond stage at the t-step.
In this study, decision stumps were used as weak learners.
A decision stump is a naive classification function in
which, for a subject with a feature vector x of peak inten-
sities, the label is predicted by observing whether a certain
peak intensity is larger than a predetermined value or not.
Accordingly, the set of weak learners is relatively large, but
all of the weak learners are literally weak, since they
respond only to a peak pattern. The set of the decision
stumps is denoted by
{dj(x) = sign(xj - b) | j = 1, ? , J, -∞ <b < ∞ }.
AdaBoost efficiently integrates the set of weak learners by
sequential minimization of the exponential loss. As a
result, the learning process of AdaBoost can be traced, and
the final classification function can be reexpressed as the
sum of the peak pattern functions Fj(x)'s, where
in which the sum of coefficients βt is referred to as the
score Sj[12]. In this way, the score Sj expresses the degree
of importance for the j-th peak in terms of contribution to
integrating a final classification function in the process of
learning algorithm.
Figure 5 (a) shows the test error of the classification result
by AdaBoost with the discrete covariates. The test error is
calculated by the test dataset with n = 50 observations sep-
arated from the training dataset. Figure 5 (b) shows the
false negative and false positive results. The results
obtained by AdaBoost with the continuous covariates are
shown in Fig. 6. These figures show typical behaviors of
AdaBoost. In Figs. 5 (b) and 6 (b), the false negatives are
much larger than the false positives. Table 1 shows the test
error of AdaBoost, where the iteration number T is
decided by 10-fold cross validation.
In Table 1, the test error with discrete covariates and that
with continuous covariates are the same.
The score Sj is calculated to consider the difference
between the obtained prediction functions. Each Sj gives
the influence of the j-th covariate for the obtained classifi-
cation function. Tables 2 and 3 show the ten highest val-
ues among Sj's in the discrete and continuous cases,
respectively. From Tables 2 and 3, the prediction func-
tions obtained using the discrete and continuous variables
are different. The result obtained using continuous covari-



F x
( ; )
f x
t t
β
t
T
( ),
β=
=∑
1

exp[{ ( )
f x
1 1
( )}].
x
i
−++
( )
2
=∑
i
yf
iiT
N
T
ββ
1
?
F x
j
( )
xb
t
t fd
jt
tj
( ),
{|}
=−
=∑
β sign