Analysis: an

Downstream statistical analysis library. Pure functions with no project-specific knowledge, organised into four subpackages by role:

Subpackage	Input	Output
compute	numpy arrays / DataFrames	arrays / dicts (no I/O)
plot	results + output_path	figure written to disk
filters	obs or var DataFrame	boolean np.ndarray mask
config	none (static)	configuration dicts

Compute

Pure statistical compute functions. No AnnData dependency (except ep_classification).

smftools.analysis.compute.autocorrelation	autocorrelation.py — NaN-aware binary autocorrelation over irregularly spaced positions.
smftools.analysis.compute.pearson	NaN-aware position × position Pearson correlation matrices.
smftools.analysis.compute.hmm_features	Interval extraction from binary HMM feature layers.
smftools.analysis.compute.ep_classification	Enhancer/promoter NDR state classification per read.
smftools.analysis.compute.ls_periodicity	ls_periodicity.py — Lomb-Scargle periodogram utilities for nucleosome analysis.
smftools.analysis.compute.dimensionality_reduction	dimensionality_reduction.py — PCA → UMAP → KNN → Leiden pipeline for per-read matrices.
smftools.analysis.compute.read_cache	Reader/writer utilities for the per-read modification matrix cache.

Plot

Figure rendering. Accepts results and an explicit output_path; writes a figure to disk.

smftools.analysis.plot.heatmaps	heatmaps.py — Pearson correlation and covariance heatmap rendering.
smftools.analysis.plot.histograms	histograms.py — Interval distribution histograms with rolling mean, peak calling, and optional Gaussian fit overlay.
smftools.analysis.plot.autocorr	autocorr.py — ACF overlay, LS periodogram, paired barplot, and metric histogram rendering.
smftools.analysis.plot.locus	locus.py — Locus map rendering for SMF accessibility data.

Filters

Boolean mask builders for obs-level and var-level (position) selection.

smftools.analysis.filters.obs_filters	Composable obs-level filters for AnnData.
smftools.analysis.filters.position_filters	position_filters.py — Genomic position selection from AnnData var coordinates.

Config

Static configuration objects; no runtime inputs.

smftools.analysis.config.hmm_histogram

Per-feature-type binning, rolling-mean, and peak-calling config for HMM feature histograms.