merquaco.data_loss.FOVDropout.detect_dropouts

static FOVDropout.detect_dropouts(transcripts: DataFrame, fovs: DataFrame, thresh: float = 0.15) → DataFrame

Compares cardinal neighbors for each FOV to detect dropout

FOVs are considered dropped if:

  • below delta threshold for number of transcripts for all 4 neighbors

  • OR below threshold for 3 neighbors and last neighbor is also a dropped FOV

Parameters:

  • transcripts (pd.DataFrame) – Transcripts table

  • fovs (pd.DataFrame) – FOVs table dataframe

  • threshold (float, optional) – Transcript count delta threshold for dropout detection

Returns:

fovs – Adds columns for each gene/FOV pair for deltas, dropout status, and if neighbors average >100 transcripts

Return type:

pd.DataFrame