Protein–protein interaction prediction 

is a field combining bioinformatics and structural biology in an

attempt to identify and catalog physical interactions between pairs or groups of proteins.
Understanding protein–protein interactions is important for the investigation of intracellular signaling
pathways, modelling of protein complex structures and for gaining insights into various biochemical
processes.

Two-hybrid screening
Two-hybrid screening (also known as yeast two-hybrid system or Y2H) is a molecular
biology technique used to discover protein-protein interactions[1] and protein-DNA interactions[2][3] by
testing for physical interactions (such as binding) between two proteins or a single protein and
a DNA molecule, respectively.

The premise behind the test is the activation of downstream reporter gene(s) by the binding of
a transcription factor onto an upstream activating sequence (UAS). For two-hybrid screening, the
transcription factor is split into two separate fragments, called the binding domain (BD) and activating
domain (AD). The BD is the domainresponsible for binding to the UAS and the AD is the domain
responsible for the activation of transcription.

Overview of two-hybrid assay, checking for interactions between two proteins, called here Bait and Prey.
A. Gal4 transcription factor gene produces two domain protein (BD and AD), which is essential for transcription of the
reporter gene (LacZ).
B,C. Two fusion proteins are prepared: Gal4BD+Bait and Gal4AD+Prey. None of them is usually sufficient to initiate the
transcription (of the reporter gene) alone.
D. When both fusion proteins are produced and Bait part of the first interact with Prey part of the second, transcription of the
reporter gene occurs.

Basic principle
The key to the two-hybrid screen is that in most eukaryotic transcription factors, the activating and binding
domains are modular and can function in close proximity to each other without direct binding.[5] This
means that even though the transcription factor is split into two fragments, it can still activate transcription
when the two fragments are indirectly connected.

The most common screening approach is the yeast two-hybrid assay.[6] This system often utilizes
a genetically engineered strain of yeast in which the biosynthesis of certain nutrients (usually amino
acids or nucleic acids) is lacking. When grown on media that lacks these nutrients, the yeast fail to
survive. This mutant yeast strain can be made to incorporate foreign DNA in the form of plasmids. In
yeast two-hybrid screening, separate bait and prey plasmids are simultaneously introduced into the
mutant yeast strain.

Plasmids are engineered to produce a protein product in which the DNA-binding domain (BD) fragment is
fused onto a protein while another plasmid is engineered to produce a protein product in which the
activation domain (AD) fragment is fused onto another protein. The protein fused to the BD may be
referred to as the bait protein, and is typically a known protein the investigator is using to identify new
binding partners. The protein fused to the AD may be referred to as the prey protein and can be either a
single known protein or a library of known or unknown proteins.
f the bait and prey proteins interact (i.e., bind), then the AD and BD of the transcription factor are indirectly
connected, bringing the AD in proximity to the transcription start site and transcription of reporter gene(s)
can occur. If the two proteins do not interact, there is no transcription of the reporter gene. In this way, a
successful interaction between the fused protein is linked to a change in the cell phenotype.[1]

To link the interaction to a change in observable phenotype, a reporter gene is provided with the
upstream activation sequence (UAS) the binding domain binds to, resulting in gene expression in
successful cases of interaction. The lacZ reporter gene will allow the highlighting of cells where the UAS-
BD-AD interaction takes place.[1] β-galactosidase, the protein product of the lacZ gene produces a blue
colouration through the metabolism of X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactoside), which allows
the experimenter to manually choose the individuals that host proteins displaying the required [level of]
interaction.

Manual differentiation of cells according to colour may be acceptable for small numbers of cells, as when
investigating a small number of proteins, but when a large library of proteins must be screened, some
automation is necessary.[2] This automation is provided by a number of genes and gene systems that
either cause the death of cells that don't host interactions, or vice versa, leaving only the cells that
express the proteins of interest.

Applications
Drug and poison discovery.
Determination of protein function.
Zinc finger protein selection for protein engineering.
Supervised learning problem
The problem of PPI prediction can be framed as a supervised learning problem. In this paradigm the
known protein interactions supervise the estimation of a function that can predict whether an interaction
exists or not between two proteins given data about the proteins (e.g., expression levels of each gene in
different experimental conditions, location information, phylogenetic profile, etc.).

Relationship to docking methods

The field of protein–protein interaction prediction is closely related to the field of protein–protein docking,
which attempts to use geometric and steric considerations to fit two proteins of known structure into a
bound complex. This is a useful mode of inquiry in cases where both proteins in the pair have known
structures and are known (or at least strongly suspected) to interact, but since so many proteins do not
have experimentally determined structures, sequence-based interaction prediction methods are especially
useful in conjunction with experimental studies of an organism's interactome.