[BiO BB] 2D Electrophoresis Gel Image in Proteomics

[John G. Hoey, Ph.D.]

Check out 2D DIGE technology from Amersham!

MyungHo Kim <bio_front at hotmail.com> wrote: 2D Electrophoresis Gel Image and Diagnosis of a Disease

The process of diagnosing a disease from the 2D gel electrophoresis
image is a challenging problem. This is due to technical difficulties of
generating reproducible images with a normalized form and the effect of
negative stain. Here, we discuss a new concept of interpreting the 2D images
and overcoming the aforementioned technical difficulties. This concept makes
use of 2D gel images of proteins in serums and we explain a way of
representing the images as vectors in order to apply machine-learning
methods, such as the support vector machine, a decision tree, and neural
network etc. (For details, see www.biofront.biz or
http://arxiv.org/abs/cs.CC/0305048)

1. Representation

I) Taking the whole image

By enumerating the whole set of numbers (densities) corresponding to each
pixel in a predetermined order, we will represent an image as a vector in a
finite dimensional Euclidean space.

II) Choosing spots

Choose a finite number, K, of conspicuous spots representing proteins and
their quantities, for example, we may take CA-1, BD-1 CA-2 and CA-3. Each of
these chosen spots will have a corresponding number, which is the sum of the
numbers assigned to each pixel consisting of the spot. Thus, the sum of each
spot will represent the relative quantity of the protein corresponding to
the spot relative to other spots. By enumerating the quantities of those
four proteins, we have a vector in the four dimensional Euclidean space.

Discussions: At a glance, in representing an electrophoresis image, the
second method seems more natural than the first one. However, though
considering the quantities of proteins looks intuitive and appealing to
biological meaning, the procedure of measuring the relative quantities of
chosen proteins may not be accurate for our purpose. On the contrary,
accepting the whole image could contain more than we realize. We all know
from a meticulous analysis that recognition of a person with a picture is
due to the human brain¡¯s ability of computing relative positions of
specific objects such as nose, eye, mouth, ears, distance between eyes. Each
pixel with its own density plays a role as a member of a whole image. Though
each pixel does not give any clue by itself, all pixels together with others
send us a concrete picture we conceive. Therefore, it is reasonable to say
that the intrinsic invariants of an image are the relative position of a
pixel with its density. The first method is about considering the whole
package of all relative positions and their densities.


2. Problems in 2D gel image and its staining methods

2D gel electrophoresis is a method that separates proteins in a
2-dimensional plane by mass and pH of proteins. As is often the case in the
most of experiments, there are two technical problems we have to compromise
so that the process of numericalization is acceptable and tolerable.

1. When the amount of a certain protein reaches a ¡°threshold¡±, the
silver stain density decreases. This phenomenon is known as the negative
staining effect.
2. Even if the experiments are performed carefully, there always will be
some variations of images. For example, in the image, the same protein will
not be in the same position relative to other proteins.

Discussions: Conceptually we need to find out a method of representation
of each serum, which is reproducible with some tolerable variations. This
attempt, as we did in the previous discussion, is feasible. Whatever
qualities we observe, estimate or sample, there are variations in
measurements or recording, since everything changes and it is impossible to
produce the equivalent results every time. For example, our heights changes
at morning and night, and blood pressures vary every hour. Likewise, there
are variations in 2D gel images even under the assumption that all the
experiments are perfectly accurate and executed the exactly same way. It
could be caused by the status of a donor of serums or experimental setting.
However, patterns of the variations seem quite consistent and, if we ponder
our heights in micrometer, then it is even natural to accept an image as it
is.



_______________________________________________
BiO_Bulletin_Board maillist - BiO_Bulletin_Board at bioinformatics.org
https://bioinformatics.org/mailman/listinfo/bio_bulletin_board



---------------------------------
Do you Yahoo!?
Free online calendar with sync to Outlook(TM).
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.bioinformatics.org/pipermail/bbb/attachments/20030603/cd26af3b/attachment.html>