Read classifications from H5 files

Reads thresholded class assignments from H5 classification files produced by the IFCB neural network classifier. Each H5 file contains:

• roi_numbers: integer vector of ROI (Region of Interest) IDs

• class_name: character vector of predicted class per ROI

• output_scores: matrix of class probabilities (classes x ROIs); the maximum score per ROI is used as the confidence value

Usage

read_h5_classifications(h5_dir, sample_ids = NULL)

Arguments

h5_dir

Directory containing .h5 files.

sample_ids

Optional character vector of sample PIDs to read. If NULL, reads all .h5 files in the directory.

Value

A data.frame with columns: sample_name, roi_number, class_name, score.