Skip to content

Visualization

The dnanet.evaluation.visualization module provides matplotlib-based plotting functions for DNA electropherogram (EPG) profiles.

Functions

from dnanet.evaluation.visualization import (
    plot_profile,
    plot_profile_marker,
    coerce_class_map,
)

plot_profile

plot_profile(signal, *, annotation=None, prediction=None, title=None, dye_colors=None, figsize=(20, 20)) — Plot a full DNA profile with optional multiclass annotations.

Args: - signal — (C, L) EPG signal data (one row per dye channel) - annotation — (C, L) class-index ground-truth annotation map - prediction — (C, L) class-index prediction map - title — Optional figure title - dye_colors — Colors for each dye channel (defaults to standard forensic dye colors) - figsize — Figure size in inches

Returns: The matplotlib Figure.

Layout: Multi-panel with signal, annotation (optional), and prediction (optional) tracks for each dye channel.

plot_profile_marker

plot_profile_marker(signal, scaler, marker_bp_range, dye_row, *, annotation=None, prediction=None, title=None, zoom=None) — Plot a single marker region of an EPG profile.

Args: - signal — (C, L) full EPG signal - scaler — (L,) scan-to-base-pair mapping array - marker_bp_range — (left_bp, right_bp) bin range of the marker - dye_row — Dye channel index for this marker - annotation — (C, L) ground-truth annotation mask - prediction — (C, L) predicted mask - title — Plot title (usually the marker name) - zoom — Optional (start, end) scan indices to zoom into

Returns: The matplotlib Figure.

coerce_class_map

coerce_class_map(data, *, signal_shape, source) — Normalize plotting labels to a 2-D class-index map.

Supports (C, L), (K, C, L), and (C, L, K) shaped arrays. Binary probability maps are thresholded at 0.5 for predictions; multiclass arrays are reduced with argmax.

Private Helpers

The following private functions support the public API:

  • _coerce_signal(signal) — Normalize signal to (num_dyes, scanpoints)
  • _plot_lines(axs, data, color, scale_to, alpha) — Plot dye channel lines
  • _plot_class_tracks(axs, class_map, lane_label) — Plot per-class mini-lanes
  • _iter_class_spans(class_row) — Return contiguous non-zero class spans
  • _plot_segmentation_mask(axs, mask, color, alpha, y_range, min_width) — Plot mask as colored tracks
  • _iter_mask_spans(mask_row, min_width) — Return contiguous positive mask spans