diffsvg.tl.StabilityResult#

class diffsvg.tl.StabilityResult(stats, ranks, similarity, top_overlap, sensitivity, configs, noise_floor, baseline, rank_by, top_k, genes)#

What check_stability found; consumed by diffsvg.pl.plot_stability.

Variables:
ranks stats,

genes x runs tables of the ranking statistic and of the rank derived from it (1 = strongest).

similarity DataFrame

runs x runs Spearman correlation between rank columns.

top_overlap DataFrame

runs x runs fraction of each run’s top top_k genes that it shares with the other run.

sensitivity DataFrame

One row per configuration that differs from the baseline in exactly one parameter, with its agreement against the baseline - i.e. the one-factor-at-a-time view of which knob actually moves the ranking.

configs DataFrame

runs x parameters table of what each run was.

noise_floor dict

Baseline-vs-reseeded-baseline agreement, as {"spearman": ..., "top_overlap": ...}. Agreement at or below this is within MCMC noise.

genes baseline, rank_by, top_k,

The baseline run’s label, and the settings the comparison used.

Attributes table#

Attributes#

stats: DataFrame#
ranks: DataFrame#
similarity: DataFrame#
top_overlap: DataFrame#
sensitivity: DataFrame#
configs: DataFrame#
noise_floor: dict#
baseline: str#
rank_by: str#
top_k: int#
genes: list#