GUI Tour

This page walks through every panel of the exo2micro GUI, top to bottom. It’s the reference you come back to when you’re not sure what a particular control does.

Todo

Add an annotated screenshot of the full GUI here, with callouts labelling each panel. Place it at docs/source/users/_images/gui_overview.png.

Input Selection

This is where you pick which samples and dyes to process.

Samples and Dyes are simple text areas — one entry per line. Processing runs every sample × dye combination.

Auto-detect scans your raw directory for sample subfolders and extracts likely dye names from the filenames. Saves typing when you have many samples.

Survey raw channels reads a small central crop from every raw TIFF and reports which RGB channels carry signal. This is a pre-flight sanity check — run it once at the start of a new dataset to confirm exo2micro’s automatic channel detection will pick the right channel for each dye.

Scale

The Scale dropdown controls how exo2micro estimates the scale factor used to subtract the pre-stain background:

Auto (Moffat fit) — the default. A Moffat profile is fit to the log-ratio distribution and the peak is used as the scale. Works well for most samples.
Auto + ratio percentile — the Moffat fit still runs, and in addition a second difference image is computed using a chosen percentile of the log-ratio distribution as the scale. A text box appears for you to enter the percentile (accepts decimals, e.g. 99.1).
Auto + manual override — the Moffat fit still runs, and in addition a second difference image is computed using an exact scale value you type in.
Auto + percentile + manual — produces all three.

See Scale Methods for guidance on when to reach for each.

Execution Options

Parallel (multiprocessing) — check this to run multiple samples concurrently. For small batches (1-3 samples) the overhead of starting workers often makes serial faster; for larger batches parallel is usually a big win.

Workers — how many parallel worker processes to run. Rule of thumb: workers × peak RAM per sample < your total RAM. Start with 4 and adjust based on what your machine can handle.

Force rerun (ignore checkpoints) — normally exo2micro skips any stage whose checkpoint already exists on disk. Checking this box re-runs everything from scratch.

From stage — where to start. Auto (resume) picks up from the latest checkpoint. Pick a specific stage when you’ve changed a parameter that only affects that stage onward. For example, if you want to try a different scale_percentile value, pick stage 4.

To stage — where to stop. Useful if you only want to regenerate the diagnostic plots (stop at 4) without touching alignment.

Show diagnostic plots inline — displays the stage-4 diagnostic plots in the notebook as each sample finishes, with short captions explaining each. Turn it off if you’re batch-processing dozens of samples and don’t want the output cell to balloon.

Advanced Parameters

Collapsed by default. When expanded, shows every tunable parameter organized into four tabs (one per pipeline stage). Each widget shows the parameter’s short abbreviation (used in filenames) and a one-line description.

Most users never need to touch these. If you do, see Parameter Reference for a full reference.

The three action buttons

▶ Run Pipeline — kicks off processing on the current sample/dye selection.

📋 Check Status — prints a checklist showing which checkpoint files exist on disk for each sample × dye combination (under the current parameter settings). Useful for seeing where you left off.

↺ Reset Params — clears any advanced parameter changes and restores defaults.

Parameter Comparison

A convenience for sweeping one parameter across several values. Pick a parameter from the dropdown, enter comma-separated values (e.g. 10, 15, 20), and click Compare. exo2micro re-runs the affected stage for each value on the first sample × dye combination and shows the results side by side.

If Save variants to disk is checked, each run’s checkpoints are written with the non-default value embedded in the filename (e.g. 02_icp_aligned_pre_bw10.tiff), so variants coexist without overwriting each other.

🔍 Zoom & Inspect

See Zoom & Inspect.

👁️ Blink Comparison

A visual tool for checking alignment quality. Load two alignment checkpoints — typically the stage-1 post-stain and either the stage-2 ICP-aligned pre-stain or the stage-3 interior-aligned pre-stain — then click the Blink: A ⇄ B toggle to flip between them at any region of interest.

Good alignment means structures barely shift when you flip. Bad alignment means things jump around.

Workflow:

Enter your sample and dye.
Pick image A (usually Post, reference, stage 1) and image B (usually Interior-aligned pre, stage 3).
Click Load. Both images are loaded as downsampled previews.
Use the Row / Col / Size sliders to frame a region of interest (a conspicuous feature in the sample is ideal).
Click the Blink: A ⇄ B toggle button repeatedly to flip between the two images. Features that stay in place are well-aligned; features that shift indicate residual error.

To diagnose where in the pipeline an alignment problem came from, compare stage 2 (ICP) against stage 3 (interior). If stage 2 is already bad, the boundary alignment failed and you want to look at boundary_width / boundary_smooth. If stage 2 is good but stage 3 is bad, SIFT matching in the interior went wrong; check interior_blur_base or interior_min_inlier_ratio.

Progress bar and output

As the pipeline runs, the progress bar tracks total tasks completed. The output area below shows real-time text from each pipeline stage plus the inline diagnostic plots (when enabled) and the final summary table.

The summary table shows the Moffat scale for every completed run, plus — when you used scale_percentile or manual_scale — extra columns for those alternative scales.