Properties

LocData, the data class for localization data, carries certain properties that describe individual (localization properties) or averaged features of the underlying localizations or groups thereof (LocData properties). In the following we provide names (or keys) for those properties.

The datatype for all keys is string. We stick to the following conventions:

• be explicit

• start with lower case letters

• use underscore

• do not use CamelCase or blanks

• use reverse notation in the sense that coordinate identifiers or identifiers of statistical functions are added in the end (position_x_mean_mean)

A list of well defined property keys used throughout locan is given by the constant: locan/constants/PropertyKey. An up-to-date description can be inspected by locan/constants/PropertyKey.index.value.description.

Localization properties:

Each localization has properties that can usually be identified in the various input (file) formats. We will use the following keys (where c stands for the coordinate x, y or z ):

• index

  localization index

• position_c

  coordinate for the c-position

• frame

  frame number in which the localization occurs

• intensity

  intensity or emission strength as estimated by the fitter

• local_background

  background in the neighborhood of localization as estimated by the fitter

• local_background_sigma

  variation of local background in terms of standard deviation

• signal_noise_ratio

  ratio between mean intensity (i.e. intensity for a single localization) and the standard deviation of local_background (i.e. local_background_sigma for a single localization)

• signal_background_ratio

  ratio between mean intensity (i.e. intensity for a single localization) and the local_background

• chi_square

  chi-square value of the fitting procedure as estimated by the fitter

• psf_sigma_c

  sigma of the fitted Gauss-function in c-dimension as estimated by the fitter

• psf_sigma

  sigma of the fitted Gauss-function - being isotropic or representing the root-mean-square of psf_sigma_c for all dimensions

• psf_width_c

  full-width-half-max of the fitted Gauss-function in c-dimension as estimated by the fitter

• psf_width

  full-width-half-max of the fitted Gauss-function - being isotropic or representing the root-mean-square of psf_width_c for all dimensions

• uncertainty_c

  localization error in c-dimension estimated by a fitter or representing a value proportional to psf_sigma_c / sqrt(intensity)

• uncertainty

  localization error for all dimensions or representing a value proportional to psf_sigma / sqrt(intensity) or representing the root-mean-square of uncertainty_c for all dimensions.

• channel

  identifier for various channels

• two_kernel_improvement

  a rapidSTORM parameter describing the improvement from two kernel fitting

• frames_number

  number of frames that contribute to a merged localization

• frames_missing

  number of frames that occurred between two successive localizations

Additional localization properties are computed by various analysis procedures including (among others):

• cluster_label

  identifier for a localization cluster

• nn_distance

  nearest-neighbor distance

• nn_distance_k

  k-nearest-neighbor distance (k can be any integer)

• colocalization_cbc

  coordinate-based colocalization value

LocData properties:

A group of localizations makes up a LocData entity for which further properties are defined.

The coordinates for a localization group are defined by their centroids.

In general, we will use the following keys for statistics of localization properties:

• property_stats

  where property is the localization property key and stats is one of the following:

  • count (number of elements)

  • min (minimum of all elements)

  • max (maximum of all elements)

  • sum (sum of all elements)

  • mean (the mean of all elements)

  • std (standard deviation of all elements)

  • sem (standard error of the mean of all elements)

For example:

• intensity_sum

  total intensity of all localizations in the group

Some properties are derived from a hull of all element positions. We provide four hulls:

1. bounding box

2. convex hull

3. alpha shape

4. oriented bounding box

From each hull a region measure (e.g. the area in 2D) and a subregion measure (e.g. the circumference in 2D) is computed.

We will use the following keys for additional properties (where c stands for the coordinate x, y or z and h stands for the corresponding hull bb, ch, as, obb):

• centroid

  tuple with mean of all localization coordinates

• localization_count

  number of localizations within a group

• region_measure_h

  area/volume (for all possible hulls)

• subregion_measure_h

  circumference/surface (for all possible hulls)

• localization_density_h

  density of localizations (for all possible hulls)

• boundary_localizations_h

  absolute number of localizations on boundary (for all possible hulls)

• boundary_localizations_ratio_h

  ratio between number of localizations on hull boundary and within hull (for all possible hulls)

• max_distance

  maximum distance between any two localizations

• inertia_moments

  inertia moments of all localizations

• orientation_obb

  angle between the x-axis and the long axis of the oriented bounding box

• orientation_im

  angle between inertia moment principal component vectors

• circularity_obb

  elongation estimated from oriented bounding box

• circularity_im

  excentricity estimated from inertia moments