In this context, in silico methods, such as Molecular Dynamics
(MD) simulations, have been used to identify the most important
residues and pathways involved in structural communication, due
to their ability to inform about how the protein behaves in the
ensemble at least on a limited timescale [1, 2, 5–7]. To extract
information about the most important residues involved in communication and possible structural communication pathways, as
well as to identify key structurally important residues and important
interactions between residues, protein structure networks (PSNs)
have been proved to be a particularly useful representation of the
protein structure and dynamics. In the following paragraph, we will
summarize how PSNs are derived and analyzed, and detail our
approach for their investigation – implemented in the PyInteraph
software [8].
1.2 Definition
of Protein Structure
Networks
Networks or graphs have been used extensively to map and analyze
relations between elements of a system in disparate fields. This
powerful model allows to keep track of and study the behavior of
complex systems in which the interaction of single elements gives
rise to emergent behaviors. A graph is a mathematical model composed of unique elements (nodes) that are connected to each other
through arcs or edges according to the existence of a specific
relation between each pair. Such networks can be represented as
square matrices called adjacency matrices, in which each row and
column corresponds to an individual node, and a value different
from zero is present in the respective position in the matrix if an
edge is present between those two nodes. A number or weight can
be associated to an edge in order to quantify the extent of their
relation. Once a system is represented as a network, analysis techniques can be applied agnostically with respect to the type of data
they encode, making many different analyses and methods available.
Not surprisingly, the network paradigm has been applied in the field
of structural biology as well, among others, in the form of protein
structure networks (PSNs), also called residue interaction networks
(RINs) or amino acid networks (AANs). Extensive reviews on what
PSNs are and how they are used in the context of structural biology
are available elsewhere [9–13] and here, we will summarize the
main concepts behind them. A PSN is a network representation
of the relations between residues in a protein, calculated either from
a single experimental structure or an ensemble of conformations,
depending on how the network is defined. The definition of what
nodes and edges correspond to in the protein structure is what
defines the network, the most common choice being to consider
the monomeric unit of a protein (i.e., the residue) as a node. Edges
are undirected, meaning they have no specified direction, and can
be defined in a number of ways depending on which property or
feature is used to measure the relation between residues, and if this
relation is detected on a single protein structure or ensemble of
154
Matteo Lambrughi et al.
(MD) simulations, have been used to identify the most important
residues and pathways involved in structural communication, due
to their ability to inform about how the protein behaves in the
ensemble at least on a limited timescale [1, 2, 5–7]. To extract
information about the most important residues involved in communication and possible structural communication pathways, as
well as to identify key structurally important residues and important
interactions between residues, protein structure networks (PSNs)
have been proved to be a particularly useful representation of the
protein structure and dynamics. In the following paragraph, we will
summarize how PSNs are derived and analyzed, and detail our
approach for their investigation – implemented in the PyInteraph
software [8].
1.2 Definition
of Protein Structure
Networks
Networks or graphs have been used extensively to map and analyze
relations between elements of a system in disparate fields. This
powerful model allows to keep track of and study the behavior of
complex systems in which the interaction of single elements gives
rise to emergent behaviors. A graph is a mathematical model composed of unique elements (nodes) that are connected to each other
through arcs or edges according to the existence of a specific
relation between each pair. Such networks can be represented as
square matrices called adjacency matrices, in which each row and
column corresponds to an individual node, and a value different
from zero is present in the respective position in the matrix if an
edge is present between those two nodes. A number or weight can
be associated to an edge in order to quantify the extent of their
relation. Once a system is represented as a network, analysis techniques can be applied agnostically with respect to the type of data
they encode, making many different analyses and methods available.
Not surprisingly, the network paradigm has been applied in the field
of structural biology as well, among others, in the form of protein
structure networks (PSNs), also called residue interaction networks
(RINs) or amino acid networks (AANs). Extensive reviews on what
PSNs are and how they are used in the context of structural biology
are available elsewhere [9–13] and here, we will summarize the
main concepts behind them. A PSN is a network representation
of the relations between residues in a protein, calculated either from
a single experimental structure or an ensemble of conformations,
depending on how the network is defined. The definition of what
nodes and edges correspond to in the protein structure is what
defines the network, the most common choice being to consider
the monomeric unit of a protein (i.e., the residue) as a node. Edges
are undirected, meaning they have no specified direction, and can
be defined in a number of ways depending on which property or
feature is used to measure the relation between residues, and if this
relation is detected on a single protein structure or ensemble of
154
Matteo Lambrughi et al.