htmd.metricdata module#

class htmd.metricdata.MetricData(dat=None, ref=None, description=None, simlist=None, fstep=0, parent=None, file=None, trajectories=None, cluster=None)#

Bases: object

Class used for storing projected trajectories, their clustering and state assignments. Objects of this class are constructed by the project methods of the other projection classes. Only construct this class if you want to load saved data.

Parameters:

  • dat (list | ndarray | None) – The projected metrics

  • ref (list | ndarray | None) – Reference indices to the simulations and frames that generated the metrics

  • description (DataFrame | None) – A description of the metrics

  • simlist (ndarray | None) – A simulation list generated by the simlist function

  • fstep (float) – Size of simulation step in ns

  • parent (MetricData | None) – The MetricData object that was used to generate this object

  • file (str | None) – Path to a saved MetricData object

  • trajectories (list | None) – A list of Trajectory objects

  • cluster (list | ndarray | None) – A list of cluster indexes for each frame in the data

trajectories#

A list of Trajectory objects

Type:

list of Trajectory objects

fstep#

Size of simulation step in ns

Type:

float

description#

A description of the metrics

Type:

pandas.DataFrame

K#

Number of clusters

Type:

int

N#

Number of frames in each cluster

Type:

numpy.ndarray

Centers#

Cluster centers (if available)

Type:

numpy.ndarray

property St#

List of the cluster assignments (state trajectory) of each trajectory.

abs2rel(absFrames)#

Convert absolute frame indexes into trajectory index-frame pairs

Useful when doing calculations on a concatenated data array of all trajectories. When you find a frame of interest you can deconcatenate the frame index to the corresponding trajectory index-frame pair.

Parameters:

absFrames (int | list | ndarray) – An absolute frame index or a list of absolute frame indexes

Returns:

pairs – A array where each row is a trajectory index-frame pair

Return type:

ndarray

Examples

>>> relidx = data.abs2rel(536)
abs2sim(absFrames)#

Converts absolute frame indexes into Sim-frame pairs

Parameters:

absFrames (int | list | ndarray) – An absolute frame index or a list of absolute frame indexes

Returns:

frames – An array of Frame objects containing the simulation object, the trajectory piece ID and the frame index.

Return type:

ndarray

Examples

>>> simframes = data.abs2sim(563)  # 563rd frame to simulation/frame pairs
property aggregateTime#

The total aggregate simulation time

Examples

>>> data.aggTime
append(other)#

Append the trajectories of another MetricData object to this one

The simulation ids of all trajectories are renumbered after appending, and any existing clustering is reset.

Parameters:

other (MetricData) – The MetricData object whose trajectories will be appended to this one

Returns:

data – This object, with the trajectories of other appended

Return type:

MetricData

bootstrap(ratio, replacement=False)#

Randomly sample a set of trajectories

Parameters:

  • ratio (float) – What ratio of trajectories to keep. e.g. 0.8

  • replacement (bool) – If we should sample with replacement

Returns:

bootdata – A new MetricData object containing only the sampled trajectories

Return type:

MetricData

Examples

>>> data = MetricSelfDistance.project(sims, 'protein and name CA')
>>> databoot = data.bootstrap(0.8)
cluster(clusterobj, mergesmall=None, batchsize=0)#

Cluster the metrics

Parameters:

  • clusterobj (ClusterMixin object) – The object of a clustering class from sklearn or with the same interface

  • mergesmall (int | None) – Clusters containing less than mergesmall conformations will be joined into their closest well-populated neighbour.

  • batchsize (int) – Batch sizes bigger than 0 will enable batching.

Examples

>>> from sklearn.cluster import MiniBatchKMeans
>>> data = MetricDistance.project(sims, 'protein and name CA', 'resname MOL')
>>> data.cluster(MiniBatchKMeans(n_clusters=1000), mergesmall=5)
combine(otherdata)#

Combines two different metrics into one by concatenating them.

Parameters:

otherdata (MetricData) – Concatenates the metrics of otherdata to the current objects metrics

Examples

>>> dataRMSD = MetricRmsd.project(sims)
>>> dataDist = MetricSelfDistance.project(sims, 'protein and name CA')
>>> dataRMSD.combine(dataDist)
copy()#

Produces a deep copy of the object

Returns:

data – A copy of the current object

Return type:

MetricData

Examples

>>> data = MetricSelfDistance.project(sims, 'protein and name CA')
>>> data2 = data.copy()
property dat#

List of the projected metrics of each trajectory.

deconcatenate(array)#

Split a concatenated data array back into per-trajectory pieces

Parameters:

array (ndarray) – A concatenated array spanning all trajectories, with as many rows as the total number of frames in the data.

Returns:

pieces – One array per trajectory, split at the trajectory boundaries.

Return type:

list

dropDimensions(drop=None, keep=None)#

Drop some dimensions of the data given their indexes

Parameters:

Examples

>>> data.dropDimensions([1, 24, 3])
>>> data.dropDimensions(keep=[2, 10])
dropFrames(idx, frames)#

Drop frames from a single trajectory of the data

Parameters:

  • idx (int) – The index of the trajectory from which to drop frames

  • frames (list | range | ndarray) – A list of frame indexes to drop

dropTraj(limits=None, multiple=None, partial=None, idx=None, keepsims=None)#

Drops trajectories based on their lengths

By default, drops all trajectories which are not of statistical mode (most common) length.

Parameters:

  • limits (list | tuple | ndarray | None) – Lower and upper limits of trajectory lengths we want to keep. e.g. [100, 500]

  • multiple (list | range | ndarray | None) – Drops trajectories whose length is not a multiple of lengths in the list. e.g. [50, 80]

  • partial (bool | None) – Not implemented yet

  • idx (list | range | ndarray | None) – A list of trajectory indexes to drop

  • keepsims (list | None) – A list of sims which we want to keep

Returns:

dropidx – The indexes of the trajectories that were dropped

Return type:

ndarray

Examples

>>> data = MetricSelfDistance.project(sims, 'protein and name CA')
>>> data.dropTraj()
>>> data.dropTraj(multiple=[100])
static fromHDF5(filename=None, h5group=None)#

Load a MetricData object from an HDF5 file or group

Parameters:

  • filename (str | None) – Path of the HDF5 file from which to load the object. Either filename or h5group must be given.

  • h5group (Group | None) – An open HDF5 group from which to read the object. Either filename or h5group must be given.

Returns:

data – The loaded MetricData object

Return type:

MetricData

load(filename)#

Load a MetricData object from disk

Parameters:

filename (str | dict | MetricData) – Path to the saved MetricData object. A dict of attributes or a legacy MetricData object is also accepted for backward compatibility.

Examples

>>> data = MetricData()
>>> data.load('./data.dat')
property map#

The description of the projected dimensions (alias for description).

property numDimensions#

The number of dimensions

Examples

>>> data.numDimensions
property numFrames#

Get the total number of frames in all trajectories

Returns:

nframes – Total number of frames in all trajectories

Return type:

int

Examples

>>> data.numFrames
property numTrajectories#

The number of trajectories

Examples

>>> data.numTrajectories
plotClusters(dimX, dimY, resolution=100, s=4, c=None, cmap='Greys', logplot=False, plot=True, save=None, data=None, levels=7)#

Plot a scatter-plot of the locations of the clusters on top of the count histogram.

Parameters:

  • dimX (int) – Index of projected dimension to use for the X axis.

  • dimY (int) – Index of projected dimension to use for the Y axis.

  • resolution (int) – Resolution of bincount grid.

  • s (float) – Marker size for clusters.

  • c (list) – Colors or indexes for each cluster.

  • cmap (matplotlib.colors.Colormap) – Matplotlib colormap for the scatter plot.

  • logplot (bool) – Set True to plot the logarithm of counts.

  • plot (bool) – If the method should display the plot

  • save (str | None) – Path of the file in which to save the figure

  • data (MetricData | None) – Optionally you can pass a different MetricData object than the one used for clustering. For example if the user wants to cluster on distances but wants to plot the centers on top of RMSD values. The new MetricData object needs to have the same simlist as this object.

  • levels (int) – Number of contour levels to draw.

plotCounts(dimX, dimY, resolution=100, logplot=False, plot=True, save=None, levels=7, cmap='viridis')#

Plots a histogram of counts on any two given dimensions.

Parameters:

  • dimX (int) – Index of projected dimension to use for the X axis.

  • dimY (int) – Index of projected dimension to use for the Y axis.

  • resolution (int) – Resolution of bincount grid.

  • logplot (bool) – Set True to plot the logarithm of counts.

  • plot (bool) – If the method should display the plot

  • save (str | None) – Path of the file in which to save the figure

  • levels (int) – Number of contour levels to draw.

  • cmap (str) – Name of the matplotlib colormap to use.

plotTrajSizes()#

Plot the lengths of all trajectories in a sorted bar plot

Examples

>>> data = MetricSelfDistance.project(sims, 'protein and name CA')
>>> data.plotTrajSizes()
property ref#

List of the reference simulation and frame indexes of each trajectory.

rel2sim(relFrames, simlist=None)#

Converts trajectory index-frame pairs into Sim-frame pairs

Parameters:

  • relFrames (list | ndarray) – An array containing in each row trajectory index and frame pairs

  • simlist (ndarray | None) – Optionally pass a different (but matching, i.e. filtered) simlist for creating the Frames.

Returns:

frames – An array of Frame objects containing the simulation object, the trajectory piece ID and the frame index.

Return type:

ndarray

Examples

>>> simframes = data.rel2sim([100, 56])  # 100th simulation frame 56
sampleClusters(clusters=None, frames=20, replacement=False, allframes=False)#

Samples frames from a set of clusters

Parameters:

  • clusters (int | list | range | ndarray | None) – A list of cluster indexes from which we want to sample. If None is given it will sample from all clusters.

  • frames (int | list | ndarray | None) – An integer with the number of frames we want to sample from each state or a list of same length as clusters which contains the number of frames we want from each of the clusters. If None is given it will return all frames.

  • replacement (bool) – If we want to sample with or without replacement.

  • allframes (bool) – Deprecated. Use frames=None instead.

Returns:

  • absframes (numpy.ndarray) – An array which contains for each state an array containing absolute trajectory frames

  • relframes (numpy.ndarray) – An array which contains for each state a 2D array containing the trajectory ID and frame number for each of the sampled frames

Examples

>>> data.sampleClusters(range(5), [10, 3, 2, 50, 1])  # Sample from first 5 clusters, 10, 3, etc frames respectively
sampleRegion(point=None, radius=None, limits=None, nsamples=20, singlemol=False)#

Samples conformations from a region in the projected space.

Parameters:

  • point (list | tuple | ndarray | None) – A point in the projected space. Undefined dimensions should have None value.

  • radius (float | None) – The radius in around the point in which to sample conformations.

  • limits (ndarray | None) – A (2, ndim) dimensional array containing the min (1st row) and max (2nd row) limits for each dimension. None values will be interpreted as no limit in that dimension, or min/max value.

  • nsamples (int) – The number of conformations to sample.

  • singlemol (bool) – If True it will return all samples within a single Molecule instead of a list of Molecules.

Returns:

  • absFrames (list) – A list of the absolute frame indexes sampled

  • relFrames (list of tuples) – A list of (trajNum, frameNum) tuples sampled

  • mols (moleculekit.molecule.Molecule or list of Molecules) – The conformations stored in a Molecule or a list of Molecules

Examples

>>> # Working with 4 dimensional data for example
>>> abs, rel, mols = data.sampleRegion(point=(0.5, 3, None, None), radius=0.1)  # Point undefined in dim 3, 4
>>> minlims = [-1, None, None, 4]  # No min limit for 2, 3 dim
>>> maxlims = [2,     3, None, 7]  # No max limit for 3 dim
>>> abs, rel, mols = data.sampleRegion(limits=np.array([minlims, maxlims]))
save(filename)#

Save a MetricData object to disk

Parameters:

filename (str) – Path of the file in which to save the object

Examples

>>> data = MetricSelfDistance.project(sims, 'protein and name CA')
>>> data.save('./data.dat')
property simlist#

List of the Sim object of each trajectory.

splitCols()#
toHDF5(filename=None, h5group=None)#

Save a MetricData object to an HDF5 file or group

Parameters:

  • filename (str | None) – Path of the HDF5 file in which to save the object. Either filename or h5group must be given.

  • h5group (Group | None) – An open HDF5 group into which to write the object. Either filename or h5group must be given.

property trajLengths#

Get the lengths of all trajectories

Returns:

lens – The lengths of all trajectories in the object

Return type:

list

Examples

>>> data.trajLengths
class htmd.metricdata.Trajectory(projection=None, reference=None, sim=None, cluster=None)#

Bases: object

Class used for storing trajectory projections, their clustering and state assignments.

Parameters:

  • projection (ndarray | None) – The projected metrics

  • reference (ndarray | None) – Reference indices to the simulations and frames that generated the metrics

  • sim (Sim | None) – A simulation object

  • cluster (ndarray | None) – A list of cluster indexes for each frame in the simulation

property cluster#

The cluster indexes for each frame in the simulation

copy()#

Produce a deep copy of the object.

Returns:

traj – A copy of the current object.

Return type:

Trajectory

dropFrames(frames)#

Drop frames from the trajectory

Parameters:

frames (list | range | ndarray) – A list of frame indexes to drop

static fromHDF5(h5group)#

Reconstruct a Trajectory from an open HDF5 group.

Return type:

Trajectory

property numDimensions#

The number of dimensions in the projection of this trajectory

property numFrames#

The number of frames in this trajectory

property projection#

The projected metrics of this simulation trajectory

property reference#

The reference indices to the simulations and frames that generated the projections

toHDF5(h5group)#

Write this trajectory’s data into an open HDF5 group.