8. Quick reference checklist

← Tutorial index

Step	Action
1	Build transcripts with required columns (Section 1).
2	Instantiate mcDETECT using the numbered parameter list (Section 2).
3	Rough then fine detection, or only detect() (Section 3).
4	profile → granule_adata (Section 4).
5	Normalize → MiniBatch k-means → heatmap → apply_manual_annotation (Section 5).
6	WT/AD density via spot windows, optional AD calibration, stats, export (Section 6).
7	Isocortex, 50×50, spot_embedding (hard), row-normalize, K-Means k = 4 (Section 7).

This document is intentionally aligned with model.py APIs and the analysis order in 3_detection.py, benchmark_subtyping.ipynb, and the Isocortex / 50×50 / hard-normalized / k = 4 branch of 7_neuropil_subdomains.ipynb.