Biology Reference
In-Depth Information
in vivo
techniques, such as yeast two-hybrid screening, and
in vitro
tech-
niques, such as coimmunoprecipitation or protein arrays (“protein
chips”). For each of these techniques, it is important to note that the
immediate result of a measurement is not just a black-and-white, quali-
tative statement on whether or not two proteins are interacting. Instead,
quantitative readouts are often available, which are then converted into
a “yes” or “no” answer using a predetermined procedure and cut-offs;
this decision may then be followed by removal of notorious contaminants
and/or rescreening. Thus, the final list of protein-protein interactions
often hides the complexity of the underlying experiment, and may not
contain all of the information from the original measurements. However,
when integrating several distinct datasets, it would be better to have this
information available: to start with the raw data for each of these datasets,
then perform a probabilistic or decision-tree-based data integration, and
only afterwards convert the final result into a set of simple qualitative
statements on who interacts with whom. While this is not yet done rou-
tinely when assembling interaction networks (and not only because the
raw data are often not available), it should probably be done more sys-
tematically in the future. In addition, many of the experimental datasets
contain intrinsic consistency signals, such as repeated measurements of
the same interaction or reverse measurements where a given protein is
the “bait” in one context and the “prey” in the other. Such intrinsic con-
sistency signals should be used as well, especially when multiple datasets
of the same type are to be integrated.
Historically, the yeast two-hybrid technology was the first method to
yield genome-wide (or near genome-wide) experimental protein interac-
tion maps.
9,10
This technology is based on artificial fusion proteins
(“hybrids”): the two proteins to be tested for an interaction are each
fused to one specifically designed protein domain — to a DNA-binding
domain in one case, and to a transcriptional transactivation domain in the
other case. Normally, these two domains occur in a single protein, form-
ing a so-called transcription factor; in two-hybrid screens, they are fused
separately to two test proteins. Only when the two test proteins are capa-
ble of forming a specific binding interaction in the nucleus of a cell (usu-
ally a
Saccharomyces cerevisiae
cell) does the reconstitution of a complete
transcription factor occur, which then drives the expression of a reporter
