Collective Variables

From PMFLib Wiki
Jump to navigation Jump to search

Navigation: Documentation / Collective Variables


Distance

Angle

Coordination

Shape

Nucleic Acids

Puckering

Special / Path

Energy

POT

Algebra

Introduction

In the framework of potential of mean force (PMF) calculations, collective variables (CVs) define the reduced space in which the free energy is calculated. The PMF is expressed as a function of a selected set of collective variables ξ, while the remaining degrees of freedom are not treated explicitly in this representation but are averaged over statistically by a computer simulation. Thus, A(ξ) represents the free energy associated with a given value of the collective variables and includes contributions from all microscopic configurations (ideal case) compatible with that value.

The vector ξ=(ξ1,ξ2,...,ξn) is the set of collective variables used in the calculation. Collective variables (CVs) are user-defined functions of the molecular geometry that describe selected, usually low-dimensional, features of a molecular system. They are used to monitor or bias processes that are difficult to sample directly in ordinary molecular dynamics simulations, such as conformational changes, ligand binding, ion transfer, or chemical transformations.

A collective variable maps the full set of atomic coordinates, x, onto one scalar quantity:

ξi=f(𝐱)...(1)

where x represents the Cartesian coordinates of the system, and ξi is the value of the collective variable. Common examples include distances, angles, torsions, coordination numbers, RMSD-like measures, and more complex path-based descriptors.

The choice and number of collective variables are critical for the efficiency and reliability of a PMF calculation. In principle, adding more collective variables can provide a more complete description of the process under study. In practice, however, the computational cost grows rapidly, often exponentially, with the dimensionality of the collective-variable space. Each additional variable increases the amount of sampling required to obtain a converged free energy surface. For this reason, PMF calculations are usually performed in a low-dimensional space, most commonly using one or two collective variables, while three or more variables are used only in special cases.

A well-chosen collective variable should capture the slow or physically relevant degrees of freedom of the process under study, distinguish important metastable states along the reaction pathway, and avoid unnecessary complexity. Poorly chosen collective variables may lead to slow convergence, hidden barriers, hysteresis, or misleading free energy profiles.

Selecting appropriate collective variables is usually a very difficult problem.

Supported CVs

PMFLib supports the following collective variables. The majority of collective variables are supported by all free-energy methods. The exception is POT because of its gradient limitation. The number of collective variables is not technically limited, but biasing methods should employ as few as possible.

Distance
Type Description Type Description
DD distance difference DIS distance between two points
MDIS minimum atom-pair distance MDISG minimum distance between two atom groups
ODIS distance between a plane and a point
OPOS absolute position in an oriented coordinate system OPOS2 absolute position in a superimposed coordinate system
ORAD radial distance in an oriented coordinate system ORAD2 radial distance in a superimposed coordinate system
POS absolute position RAD radial distance
Angle
Type Description Type Description
ANG angle defined by three points ANG2 angle between two vectors defined by four points
AXANG angle between two principal axes AXANG2 angle between two reference axes
DIH dihedral angle defined by four points DIH2 dihedral angle defined by three vectors and six points
PANG angle between two planes PVANG angle between a plane and a vector
Coordination Numbers
Type Description Type Description
CNFF atom-atom coordination number calculated using a Fermi-like function CNGFF point-atom coordination number calculated using a Fermi-like function
CNSW atom-atom coordination number calculated using a switching function CNGSW point-atom coordination number calculated using a switching function
CNRF atom-atom coordination number calculated using a rational function CNGRF point-atom coordination number calculated using a rational function
Shape
Type Description Type Description
ACYL acylindricity PMGTD difference between two principal moments of the gyration tensor
ASPH asphericity PMOGT principal moment of the gyration tensor
SANIS relative shape anisotropy RMSDT root-mean-square deviation from a target structure
RGYR radius of gyration PLANE root-mean-square deviation from a plane
EVEC projection onto an essential vector
Nucleic Acids
Type Description Type Description
NALBPP local base-pair parameters NALSTP local base-pair step parameters
NASBPP simple base-pair parameters NASSTP simple base-pair step parameters
NASBPPOLD legacy simple base-pair parameters NASSTPOLD legacy simple base-pair step parameters
NASBO simple base-pair opening NAPBO primitive base-pair opening
NABEND nucleic acid bending
Ring Puckering
Type Description Type Description
PUCK6Q total puckering amplitude of a six-membered ring PUCK5Q total puckering amplitude of a five-membered ring
PUCK6P puckering azimuthal angle of a six-membered ring PUCK5P puckering azimuthal angle of a five-membered ring
PUCK6T puckering polar angle of a six-membered ring
Special / Path
Type Description Type Description
PTPATHS path-progress coordinate for point-based paths PTPATHS2P path-progress coordinate along a line generated from two points
PTPATHS2PN path-progress coordinate along bidirectional line from origin + direction PTPATHZ path-distance coordinate for point-based paths
PATHS path-progress coordinate PATHZ path-distance coordinate
WORMPOS parametric position of an axle inside a macrocycle WORMANG orientation of an axle inside a macrocycle
Energy
Type Description Type Description
POT potential energy
Algebra
Type Description Type Description
ADD sum of two collective variables SUB difference between two collective variables
MULT product of two collective variables DIV ratio of two collective variables
FSWITCH Fermi-like switching function RSWITCH rational switching function

Specification

The CV definition can be provided either directly in the input file or in an external file. The selected mode depends on the configuration in the [files] section.

If the value of the fcvsdef key specifies the name of an input-file group, for example a string beginning with the "{" character, the CV definition is read from this group in the input file. Otherwise, the value of fcvsdef is interpreted as the name of an external file containing the CV definition. In this case, the CV definition must be placed in the {MAIN} group, which is implicit and may be omitted from the file.

A single group or file may define multiple collective variables. Each collective variable is specified in a separate section. The section name determines the CV type. Each section must contain the name key, which defines a unique name of the CV, together with additional keys specific to the selected CV type.

Extra documentation