The most important data structure in MDAnalysis is the AtomGroup, which contains Atom instances.
A Universe is the user-visible entry point and collects all information needed to analyze a structure or a whole trajectory.
Segments and residues are a way to refer to a collection of atoms. By convention, a Residue is a single amino acid, or a water molecule, ion, or ligand. A Segment is a collection of residues such as a whole protein or a chain in a protein or all the water in the system.
A Universe contains Segments, which contain Residues, which contain Atoms; all containers are derived from AtomGroup, and thus one can always analyze them as a collection of atoms, independent of the hierarchical level.
Each Atom can only belong to a single Residue, and a Residue belongs to one specific Segment. This hierarchy can be described as
Segment > Residue > Atom
Depending on the use case, it can be more convenient to access data on, for instance, the basis of residues than atoms, or to write out individual chains (segments) of a protein. MDAnalysis simply provides access to these groupings and keeps track of where an atom belongs. Each object provides three attributes (atoms, residues or residue, segments or segment) that give access to the tiers in the hierarchy that the object belongs to.
The MDAnalysis Universe contains all the information describing the system.
The system always requires a topology file — in the simplest case just a list of atoms. This can be a CHARMM/NAMD PSF file or a simple coordinate file with atom informations such as PDB, Gromacs GRO, or CHARMM CRD. See Table of Supported topology formats for what kind of topologies can be read.
A trajectory provides coordinates; the coordinates have to be ordered in the same way as the list of atoms in the topology. A trajectory can be a single frame such as a PDB, CRD, or GRO file, or it can be a MD trajectory (in CHARMM/NAMD/LAMMPS DCD, Gromacs XTC/TRR, or generic XYZ format). See Table of Supported coordinate formats for what can be read as a “trajectory”.
As a special case, when the topology is a PDB, GRO or CRD file then the coordinates are immediately loaded from the “topology” file unless a trajectory is supplied.
Examples for setting up a universe:
u = Universe(topology, trajectory) # read system from file(s)
u = Universe(pdbfile) # read atoms and coordinates from PDB or GRO
u = Universe(topology, [traj1, traj2, ...]) # read from a list of trajectories
Load new data into a universe (replaces old trajectory and does not append):
u.load_new(trajectory) # read from a new trajectory file
Select atoms, with syntax similar to CHARMM (see AtomGroup.selectAtoms for details):
u.selectAtoms(...)
Attributes:
Note
If atom attributes such as element, mass, or charge are not explicitly provided in the topology file then MDAnalysis tries to guess them (see MDAnalysis.topology.tables). This does not always work and if you require correct values (e.g. because you want to calculate the center of mass) then you need to make sure that MDAnalysis gets all the information needed. Furthermore, the list of bonds is only constructed when provided in the topology and never guessed (see Issue 23).
Changed in version 0.7.5: Can also read multi-frame PDB files with the PrimitivePDBReader.
Initialize the central MDAnalysis Universe object.
Arguments : |
|
---|
This routine tries to do the right thing:
If a pdb/gro file is provided instead of a psf and no coordinatefile then the coordinates are taken from the first file. Thus you can load a functional universe with
u = Universe('1ake.pdb')
If you want to specify the coordinate file format yourself you can do so using the format keyword:
u = Universe('1ake.ent1', format='pdb')
If only a topology file without coordinate information is provided one will have to load coordinates manually using Universe.load_new(). The file format of the topology file can be explicitly set with the topology_format keyword.
Changed in version 0.7.4: New topology_format and format parameters to override the file format detection.
Reference to current timestep and coordinates of universe.
The raw trajectory coordinates are Universe.coord._pos, represented as a numpy.float32 array.
Because coord is a reference to a Timestep, it changes its contents while one is stepping through the trajectory.
Note
In order to access the coordinates it is probably better to use the AtomGroup.coordinates() method; for instance, all coordinates of the Universe as a numpy array: Universe.atoms.coordinates().
Current dimensions of the unitcell
Load coordinates from filename, using the suffix to detect file format.
Arguments : |
|
---|---|
Returns : | (filename, trajectory_format) or None if filename == None |
Raises : | TypeError if trajectory format can not be determined or no appropriate trajectory reader found |
Select atoms using a CHARMM selection string.
Returns an AtomGroup with atoms sorted according to their index in the psf (this is to ensure that there aren’t any duplicates, which can happen with complicated selections).
Subselections can be grouped with parentheses.
>>> universe.selectAtoms("segid DMPC and not ( name H* or name O* )")
<AtomGroup with 3420 atoms>
Note
If exact ordering of atoms is required (for instance, for angle() or dihedral() calculations) then one supplies selections separately in the required order. Also, when multiple AtomGroup instances are concatenated with the + operator then the order of Atom instances is preserved and duplicates are not removed.
See also
Selection Commands for further details and examples.
The selection parser understands the following CASE SENSITIVE keywords:
Simple selections
- protein, backbone, nucleic, nucleicbackbone
- selects all atoms that belong to a standard set of residues; a protein is identfied by a hard-coded set of residue names so it may not work for esoteric residues.
- segid seg-name
- select by segid (as given in the topology), e.g. segid 4AKE or segid DMPC
- resid residue-number-range
- resid can take a single residue number or a range of numbers. A range consists of two numbers separated by a colon (inclusive) such as resid 1:5. A residue number (“resid”) is taken directly from the topology.
- resnum resnum-number-range
- resnum is the canonical residue number; typically it is set to the residue id in the original PDB structure.
- resname residue-name
- select by residue name, e.g. resname LYS
- name atom-name
- select by atom name (as given in the topology). Often, this is force field dependent. Example: name CA (for Cα atoms) or name OW (for SPC water oxygen)
- type atom-type
- select by atom type; this is either a string or a number and depends on the force field; it is read from the topology file (e.g. the CHARMM PSF file contains numeric atom types). It has non-sensical values when a PDB or GRO file is used as a topology.
- atom seg-name residue-number atom-name
- a selector for a single atom consisting of segid resid atomname, e.g. DMPC 1 C2 selects the C2 carbon of the first residue of the DMPC segment
Boolean
- not
- all atoms not in the selection, e.g. not protein selects all atoms that aren’t part of a protein
- and, or
- combine two selections according to the rules of boolean algebra, e.g. protein and not (resname ALA or resname LYS) selects all atoms that belong to a protein, but are not in a lysine or alanine residue
Geometric
- around distance selection
- selects all atoms a certain cutoff away from another selection, e.g. around 3.5 protein selects all atoms not belonging to protein that are within 3.5 Angstroms from the protein
- point x y z distance
- selects all atoms within a cutoff of a point in space, make sure coordinate is separated by spaces, e.g. point 5.0 5.0 5.0 3.5 selects all atoms within 3.5 Angstroms of the coordinate (5.0, 5.0, 5.0)
- prop [abs] property operator value
- selects atoms based on position, using property x, y, or z coordinate. Supports the abs keyword (for absolute value) and the following operators: <, >, <=, >=, ==, !=. For example, prop z >= 5.0 selects all atoms with z coordinate greater than 5.0; prop abs z <= 5.0 selects all atoms within -5.0 <= z <= 5.0.
Connectivity
- byres selection
- selects all atoms that are in the same segment and residue as selection, e.g. specify the subselection after the byres keyword
Index
- bynum index-range
- selects all atoms within a range of (1-based) inclusive indices, e.g. bynum 1 selects the first atom in the universe; bynum 5:10 selects atoms 5 through 10 inclusive. All atoms in the MDAnalysis.Universe are consecutively numbered, and the index runs from 1 up to the total number of atoms.
Changed in version 0.7.4: Added resnum selection.
Reference to trajectory reader object containing trajectory data.
A group of atoms.
ag = universe.selectAtoms(atom-list)
The AtomGroup contains a list of atoms; typically, a AtomGroup is generated from a selection. It is build from any list-like collection of Atom instances. It is also possible to create an empty AtomGroup from an empty list.
An AtomGroup can be indexed and sliced like a list:
ag[0], ag[-1]
will return the first and the last Atom in the group whereas the slice
ag[0:6:2]
returns every second element, corresponding to indices 0, 2, and 4.
It also supports “advanced slicing” when the argument is a numpy.ndarray or a list:
aslice = [0, 3, -1, 10, 3]
ag[aslice]
will return a new AtomGroup containing (ag[0], ag[3], ag[-1], ag[10], ag[3]).
Note
AtomGroups originating from a selection are sorted and duplicate elements are removed. This is not true for AtomGroups produced by slicing. Thus slicing can be used when the order of atoms is crucial (for instance, in order to define angles or dihedrals).
Atoms can also be accessed in a Pythonic fashion by using the atom name as an attribute. For instance,
ag.CA
will provide a AtomGroup of all CA atoms in the group. These instant selector attributes are auto-generated for each atom name encountered in the group.
Note
The name-attribute instant selector access to atoms is mainly meant for quick interactive work. Thus it either returns a single Atom if there is only one matching atom, or a new AtomGroup for multiple matches. This makes it difficult to use the feature consistently in scripts but it is much better for interactive work.
References for analysis methods
[Dima2004] | (1, 2) Dima, R. I., & Thirumalai, D. (2004). Asymmetry in the shapes of folded and denatured states of proteins. J Phys Chem B, 108(21), 6564-6570. doi:10.1021/jp037128y |
Changed in version 0.7.6: An empty AtomGroup can be created and no longer raises a NoDataError.
Rebuild all AtomGroup caches.
A number of lists and attributes are cached. These caches are lazily built the first time they are needed. When editing the topology it might happen that not all caches were synced properly (even though that this is supposed to happen eventually). In this case the user can manually force a complete cache rebuild.
Currently the following caches are used:
New in version 0.7.5.
Align principal axis with index axis with vector.
Arguments : |
|
---|
To align the long axis of a channel (the first principal axis, i.e. axis = 0) with the z-axis:
u.atoms.align_principalAxis(0, [0,0,1])
u.atoms.write("aligned.pdb")
Returns the angle in degrees between atoms 0, 1, 2.
Angle between atoms 0 and 2 with apex at 1:
2
/
/
1------0
New in version 0.7.3.
Asphericity.
See [Dima2004] for background information.
New in version 0.7.7.
AtomGroup of all atoms in this group
The AtomGroup itself; provided for unified access across the hierarchy. Use _atoms if you really only need a list of individual Atom instances.
Return the bounding box of the selection.
The lengths A,B,C of the orthorhombic enclosing box are
L = AtomGroup.bbox()
A,B,C = L[1] - L[0]
Returns : | [[xmin, ymin, zmin], [xmax, ymax, zmax]] |
---|
New in version 0.7.2.
Crystallographic B-factors (from PDB) in A**2.
Deprecated since version 0.8: This managed attribute will become a method in 0.8 in order to provide a unified interface to querying properties: AtomGroup.bfactors will become AtomGroup.bfactors()
Returns the distance between atoms in a 2-atom group.
Distance between atoms 0 and 1:
0---1
New in version 0.7.3.
Return the bounding sphere of the selection.
The sphere is calculated relative to the centre of geometry.
Returns : | (R, [xcen,ycen,zcen]) |
---|
New in version 0.7.3.
Center of geometry (also known as centroid) of the selection.
Center of mass of the selection.
Center of geometry (also known as centroid) of the selection.
Array of partial charges of the atoms (as defined in the topology)
NumPy array of the coordinates.
See also
Deprecated since version 0.7.6: In new scripts use AtomGroup.get_positions() preferrably.
Calculate the dihedral angle in degrees.
Dihedral angle around axis connecting atoms 1 and 2 (i.e. the angle between the planes spanned by atoms (0,1,2) and (1,2,3)):
3
|
1-----2
/
0
New in version 0.7.0.
Calculate the dihedral angle in degrees.
Deprecated since version 0.7.3: Do not use this method directly; instead simply use dihedral() which is set up to use dihedral_numpy().
Calculate the dihedral angle in degrees.
Deprecated since version 0.7.3: This is a slower implementation (without numpy) and will be removed. Use dihedral_numpy() instead (or simply dihedral() which is set up to use dihedral_numpy().)
Dimensions of the Universe to which the group belongs, at the current time step.
Forces on the atoms in the AtomGroup.
The forces can be changed by assigning an array of the appropriate shape, i.e. either Nx3 to assign individual force or 3, to assign the same force to all atoms (e.g. ag.forces = array([0,0,0]) will set all forces to (0.,0.,0.)).
For more control use the get_forces() and set_forces() methods.
New in version 0.7.7.
Get a NumPy array of the atomic forces (if available). Currently only supported for Gromacs .trr trajectories.
Keywords : |
|
---|
Forces can also be directly obtained from the attribute forces.
Forces can be directly set with set_forces() or by assigning to forces.
New in version 0.7.7.
Get a NumPy array of the coordinates.
Keywords : |
|
---|
Coordinates can also be directly obtained from the attribute positions.
Coordinates can be directly set with set_positions() or by assigning to positions.
This method is identical with coordinates() but named differently for symmetry with with set_positions().
New in version 0.7.6.
NumPy array of the velocities.
Raises a NoDataError if the underlying Timestep does not contain _velocities.
See also AtomGroup.set_velocities() and attribute access through AtomGroup.velocities.
New in version 0.7.6.
Returns the improper dihedral between 4 atoms.
The improper dihedral is calculated in the same way as the proper dihedral(): The angle between the planes formed by atoms (0,1,2) and (1,2,3).
Note
Only makes sense for a AtomGroup with exactly 4 Atom; anything else will raise a ValueError. The interpretation of the angle as an “improper” solely depends on the selection of atoms and thus the user input!
New in version 0.7.3.
Array of all Atom.number in the group.
These indices are 0-based and can be used to directly index Universe.atoms or the coordinate array MDAnalysis.coordinates.base.Timestep._pos.
Array of atomic masses (as defined in the topology)
Tensor of inertia as 3x3 NumPy array.
Returns a list of atom names.
Total number of atoms in the group
Total number of residues in the group
Total number of segments in the group
Coordinates of the atoms in the AtomGroup.
The positions can be changed by assigning an array of the appropriate shape, i.e. either Nx3 to assign individual coordinates or 3, to assign the same coordinate to all atoms (e.g. ag.positions = array([0,0,0]) will move all particles to the origin).
For more control use the get_positions() and set_positions() methods.
New in version 0.7.6.
Calculate the principal axes from the moment of inertia.
e1,e2,e3 = AtomGroup.principalAxes()
The eigenvectors are sorted by eigenvalue, i.e. the first one corresponds to the highest eigenvalue and is thus the first principal axes.
Returns : | numpy.array v with v[0] as first, v[1] as second, and v[2] as third eigenvector. |
---|
Array of atomic radii (as defined in the PQR file)
Radius of gyration.
Returns a list of residue numbers.
Read-only list of Residue objects.
A ResidueGroup of all residues that contain atoms in this group.
Returns a list of residue names.
Returns a list of canonical residue numbers.
New in version 0.7.4.
Apply a rotation matrix R to the selection’s coordinates.
AtomGroup.rotate(R)
R is a 3x3 orthogonal matrix that transforms x –> x’:
x’ = R.x
Apply a rotation to the selection’s coordinates.
AtomGroup.rotateby(angle,axis[,point])
The transformation from current coordinates x to new coordinates x’ is
x’ = R.(x-p) + p
where R is the rotation by angle around the axis going through point p.
Arguments : |
|
---|---|
Returns : | The 4x4 matrix which consists of the rotation matrix M[:3,:3] and the translation vectort M[:3,3]. |
Returns a list of segment ids (=segment names).
Read-only list of Segment objects.
A SegmentGroup of all segments that contain atoms in this group.
Selection of atoms using the MDAnalysis selection syntax.
AtomGroup.selectAtoms(selection[,selection[,...]])
See also
Set the atom bfactor to float bfactor for all atoms in the AtomGroup.
New in version 0.7.4.
Set the partial charge to float charge for all atoms in the AtomGroup.
New in version 0.7.4.
Set the forces for all atoms in the group.
Arguments : |
|
---|---|
Keywords : |
|
Note
If the group contains N atoms and force is a single point (i.e. an array of length 3) then all N atom positions are set to force (due to NumPy’s broadcasting rules), as described for forces.
See also get_forces() and attribute access through forces.
New in version 0.7.7.
Set the atom mass to float mass for all atoms in the AtomGroup.
New in version 0.7.4.
Set the atom name to string name for all atoms in the AtomGroup.
New in version 0.7.4.
Set the positions for all atoms in the group.
Arguments : |
|
---|---|
Keywords : |
|
Note
If the group contains N atoms and coord is a single point (i.e. an array of length 3) then all N atom positions are set to coord (due to NumPy’s broadcasting rules), as described for positions.
See also get_positions() and attribute access through positions.
New in version 0.7.6.
Set the atom radius to float radius for all atoms in the AtomGroup.
New in version 0.7.4.
Set the resid to integer resid for all atoms in the AtomGroup.
Note
Only changes the resid but does not create combined Residue objects or split them.
New in version 0.7.4.
Changed in version 0.7.5: Also changes the residues.
Set the resname to string resname for all atoms in the AtomGroup.
New in version 0.7.4.
Changed in version 0.7.5: Also changes the residues.
Set the resnum to resnum for all atoms in the AtomGroup.
New in version 0.7.4.
Changed in version 0.7.5: Also changes the residues.
Set the segid to segid for all atoms in the AtomGroup.
If buildsegments is set to False for performance reasons then one needs to run Universe._build_segments() in order to update the list of Segment instances and regenerate the segid instant selectors.
New in version 0.7.4.
Set the atom type to atype for all atoms in the AtomGroup.
New in version 0.7.4.
Assign the velocities v to the timestep.
Raises a NoDataError if the underlying Timestep does not contain _velocities.
See also AtomGroup.get_velocities() and AtomGroup.velocities for attribute access.
New in version 0.7.6.
Shape parameter.
See [Dima2004] for background information.
New in version 0.7.7.
Sum of all partial charges (must be defined in topology).
Total mass of the selection (masses are taken from the topology or guessed).
Apply homogenous transformation matrix M to the coordinates.
The matrix M must be a 4x4 matrix, with the rotation in M[:3,:3] and the translation in M[:3,3].
The rotation is applied before the translation:
x' = R.x + t
Apply translation vector t to the selection’s coordinates.
>>> AtomGroup.translate(t)
>>> AtomGroup.translate((A, B))
The method applies a translation to the AtomGroup from current coordinates x to new coordinates x’:
x' = x + t
The translation can also be given as a tuple of two MDAnalysis objects such as two selections (selA, selB), i.e. two AtomGroup, or two Atom. The translation vector is computed as the difference between the centers of geometry (centroid) of B and A:
t = B.centroid() - A.centroid()
Temporary Timestep that contains the selection coordinates.
A Timestep instance, which can be passed to a trajectory writer.
If ts is modified then these modifications will be present until the frame number changes (which typically happens when the underlying trajectory frame changes).
It is not possible to assign a new Timestep to the AtomGroup.ts attribute; change attributes of the object.
The universe to which the atoms belong (read-only).
NumPy array of the velocities of the atoms in the group.
If the trajectory does not contain velocity information then a NoDataError is raised.
New in version 0.7.5.
Deprecated since version 0.7.6: In 0.8 this will become an attribute! You can already use get_velocities() and set_velocities().
Write AtomGroup to a file.
AtomGroup.write(filename[,format])
Keywords : |
|
---|
Write AtomGroup selection to a file to be used in another programme.
Keywords : |
|
---|
A class representing a single atom.
Atom is the basic building block of all larger data structures in MDAnalysis, in particular of the AtomGroup.
An Atom is typically generated by a topology reader from MDAnalysis.topology.
For performance reasons, only a predefined number of attributes are included (and thus it is not possible to add attributes “on the fly”; they have to be included in the class definition).
atom number
name of the segment
residue number
canonical residue number as, for instance, used in the original PDB file
New in version 0.7.4.
residue name
string, short name
string or number (from force field), describing the atom type; stored as a string.
Changed in version 0.7.6: The Atom.type attribute is always stored as a string.
float, in atomic mass units (u)
float, in electron charges (e)
Born-radius for electrostatic calculations. (Only if read from a PQR file with the PQRReader.)
deprecated, use position, get the Current cartesian coordinates of the atom.
Get the current cartesian coordinates of the atom. @return: a numpy 1x3 array
a pointer back to the Universe
Current velocity of the atom.
A NoDataError is raised if the trajectory does not contain velocities.
New in version 0.7.5.
A group of atoms corresponding to a residue.
Using a atom name as attribute returns the matching atom (a Atom instance), i.e. r.name. Example:
>>> from MDAnalysis.tests.datafiles import PSF,DCD
>>> u = Universe(PSF,DCD)
>>> print(u.s4AKE.r1.CA) # C-alpha of M1
< Atom 5: name 'CA' of type '22' of resname 'MET', resid 1 and segid '4AKE'>
r['name'] or r[id] - returns the atom corresponding to that name
Data : |
|
---|
Changed in version 0.7.4: Added Residue.resnum attribute and resnum keyword argument.
AtomGroup corresponding to the chi1 sidechain dihedral N-CA-CB-CG.
Returns : | 4-atom selection in the correct order. If no CB and/or CG is found then this method returns None. |
---|
New in version 0.7.5.
AtomGroup corresponding to the omega protein backbone dihedral CA-C-N’-CA’.
omega describes the -C-N- peptide bond. Typically, it is trans (180 degrees) although cis-bonds (0 degrees) are also occasionally observed (especially near Proline).
Returns : | 4-atom selection in the correct order. If no C’ found in the previous residue (by resid) then this method returns None. |
---|
AtomGroup corresponding to the phi protein backbone dihedral C’-N-CA-C.
Returns : | 4-atom selection in the correct order. If no C’ found in the previous residue (by resid) then this method returns None. |
---|
AtomGroup corresponding to the psi protein backbone dihedral N-CA-C-N’.
Returns : | 4-atom selection in the correct order. If no N’ found in the following residue (by resid) then this method returns None. |
---|
A group of residues.
Using a atom name as attribute returns a list of all atoms (a AtomGroup) of the same name. Example:
>>> from MDAnalysis.tests.datafiles import PSF,DCD
>>> u = Universe(PSF,DCD)
>>> print(u.s4AKE.MET.CA) # C-alpha of all Met
<AtomGroup with 6 atoms>
Data : | ResidueGroup._residues |
---|
Initialize the ResidueGroup with a list of Residue instances.
A group of residues corresponding to one segment of the topology.
The attribute rN returns the N-th residue Residue of the segment (numbering starts at N=1). Example:
>>> from MDAnalysis.tests.datafiles import PSF,DCD
>>> u = Universe(PSF,DCD)
>>> print(u.s4AKE.r1)
<Residue 'MET', 1>
Using a residue name as attribute returns a list of all residues (a ResidueGroup) of the same name. Example:
>>> from MDAnalysis.tests.datafiles import PSF,DCD
>>> u = Universe(PSF,DCD)
>>> print(u.s4AKE.CYS)
<ResidueGroup [<Residue 'CYS', 77>]>
>>> print(u.s4AKE.MET)
<ResidueGroup [<Residue 'MET', 1>, <Residue 'MET', 21>, <Residue 'MET', 34>, <Residue 'MET', 53>, <Residue 'MET', 96>, <Residue 'MET', 174>]>
Data : | Segment.name is the segid from the topology or the chain identifier when loaded from a PDB |
---|
Initialize a Segment with segid name from a list of Residue instances.
Segment id (alias for Segment.name)
Segment id (alias for Segment.name)
A group of segments.
Using a segid as attribute returns the segment. Because of python language rule, any segid starting with a non-letter character is prefixed with ‘s’, thus ‘4AKE’ –> ‘s4AKE’.
Example:
>>> from MDAnalysis.tests.datafiles import PSF,DCD
>>> u = Universe(PSF,DCD)
>>> print(u.atoms.segments.s4AKE) # segment 4AKE
<AtomGroup with 3314 atoms>
Indexing the group returns the appropriate segment.
Data : | SegmentGroup._segments |
---|
Initialize the SegmentGroup with a list of Segment instances.
Return a universe from the input arguments.
If the first argument is a universe, just return it:
as_universe(universe) --> universe
Otherwise try to build a universe from the first or the first and second argument:
asUniverse(PDB, **kwargs) --> Universe(PDB, **kwargs)
asUniverse(PSF, DCD, **kwargs) --> Universe(PSF, DCD, **kwargs)
asUniverse(*args, **kwargs) --> Universe(*args, **kwargs)
Returns : | an instance of Universe |
---|
Raised when a atom selection failed.
Warning indicating a possible problem with a selection.
Raised when empty input is not allowed or required data are missing.