Prediction of Events (prob_para)

  • empty_prob (double | advanced parameter) — Probability of an event (0 to 1) if no training data is provided.

  • mitosis_inference_option (string) — Method of mitosis inference.

    Option

    Mother needs to have high

    probabilities of being mitotic?

    Daughters need to have high

    probabilities of being newly born?
    ‘all’ or ‘both’ Yes Yes
    ‘before’ Yes No
    ‘after’ No Yes
    ‘none’ No No

    Probabilities of “No” events will be modified to 0.5, such that these events’ states will not affect the probability of cell lineages.

  • migration_option (string) — Method of migration probability calculation.

    Option Consider Migration Distance? Consider Ellipse Similarity?
    ‘similarity’ Yes Yes
    ‘distance’ Yes No

    For ‘similarity’, the migration probability is calculated by

    \[\begin{split}& P\left(D_{i,t}\rightarrow D_{j,t+1}\right) \\ = & \frac{P_{sim}\left(D_{i,t}, D_{j,t+1}\right) P_{mig}\left(D_{i,t}, D_{j,t+1}\right)}{P_{sim}\left(D_{i,t}, D_{j,t+1}\right) P_{mig}\left(D_{i,t}, D_{j,t+1}\right) + \left[1-P_{sim}\left(D_{i,t}, D_{j,t+1}\right)\right]P_{nonmig}}\end{split}\]

    where \(D_{i,t}\) and \(D_{j,t+1}\) refer to the i-th ellipse in Frame t and j-th ellipse in Frame t+1, respectively; \(P_{sim}\left(D_{i,t}, D_{j,t+1}\right)\) is the probability that \(D_{i,t}\) and \(D_{j,t+1}\) represent the same cell; \(P_{mig}\left(D_{i,t}, D_{j,t+1}\right)\) is the probability that a cell migrates from the position of \(D_{i,t}\) to the position of \(D_{j,t+1}\) by random walk; and \(P_{nonmig}\) is a constant null probability for migration.

    For ‘distance’, the migration probability equals to \(P_{mig}\left(D_{i,t}, D_{j,t+1}\right)\).

  • migration_speed (double or string) — Method of migration speed inference.

    Defined as the standard deviation of random walk in one direction and one frame. Available options:

    Option Description
    ‘global’ Inference of one migration speed for all cells.
    ‘time’ Time-dependent inference of migration speeds.
    ‘density’ Local cell density-dependent inference of migration speeds.
    Number A custom migration speed for all cells.

  • max_migration_dist_fold (double | advanced parameter) — Maximal distance (in folds of the migration speed) a cell can migrate in a frame.

  • migration_inference_resolution (double | advanced parameter) — ‘time’ and ‘density’ only. Resolution of time (in frames) or cell density (in the number of cells) for migration speed inference.

  • migration_inference_min_samples (integer | advanced parameter) — ‘time’ and ‘density’ only. Minimal number of training samples at each time/cell density for migration speed inference.

  • prob_nonmigration (double | advanced parameter) — Null probability (0 to 1) of migration.

  • min_inout_prob (double | advanced parameter) — Minimal probability (0 to 1) to migrate in/out of the field of view.

  • max_migration_time (integer | advanced parameter) — Maximal number of frames in a migration.

    max_migration_time-1 equals to the maximal number of frames a track can skip at once.

    Note that local track correction is not optimized for tracks skipping any frames. Error rates might be high.