Visualization¶

Sequence logos, spatial model plots, and regression diagnostic figures.

Dependencies

All functions require matplotlib. The plot_pssm_logo function uses logomaker if available, falling back to a bar-chart representation.

pyprego.visualization ¶

Visualization functions for motif logos and regression diagnostics.

Mirrors plot-logo.R and plot-regression.R from the R prego package. Uses matplotlib as the base; logomaker is optional for sequence logos. All matplotlib imports are lazy so pyprego works without matplotlib installed.

plot_pssm_logo ¶

plot_pssm_logo(pssm: DataFrame, *, ax: Axes | None = None, title: str | None = None, method: str = 'bits') -> matplotlib.axes.Axes

Plot a sequence logo for a PSSM.

Tries to use logomaker for high-quality logos. If logomaker is not installed, falls back to a stacked bar chart rendered with plain matplotlib.

PARAMETER	DESCRIPTION
`pssm`	PSSM DataFrame with columns `pos`, `A`, `C`, `G`, `T`. TYPE: `DataFrame`
`ax`	Axes to plot on. If `None`, a new figure is created. TYPE: `Axes \| None` DEFAULT: `None`
`title`	Optional title for the plot. TYPE: `str \| None` DEFAULT: `None`
`method`	`"bits"` (default) scales letters by information content; `"probability"` scales by raw probability. TYPE: `str` DEFAULT: `'bits'`

RETURNS	DESCRIPTION
`Axes`	The axes with the logo drawn.

plot_spat_model ¶

plot_spat_model(spat: DataFrame, *, ax: Axes | None = None, title: str | None = 'Spatial model') -> matplotlib.axes.Axes

Plot a spatial model as a line-and-point chart.

Mirrors plot_spat_model() from R which uses geom_line + geom_point.

PARAMETER	DESCRIPTION
`spat`	Spatial model DataFrame with columns `bin` and `spat_factor`. TYPE: `DataFrame`
`ax`	Axes to plot on. If `None`, a new figure is created. TYPE: `Axes \| None` DEFAULT: `None`
`title`	Optional title. TYPE: `str \| None` DEFAULT: `'Spatial model'`

RETURNS	DESCRIPTION
`Axes`	The axes with the plot.

plot_regression_prediction ¶

plot_regression_prediction(pred: ndarray, response: ndarray, *, ax: Axes | None = None, point_size: float = 0.5, alpha: float = 1.0, title: str | None = 'Regression prediction') -> matplotlib.axes.Axes

Scatter plot of predicted vs observed response.

Mirrors the R plot_regression_prediction which shows response on x-axis and prediction on y-axis, annotated with R-squared and r.

PARAMETER	DESCRIPTION
`pred`	Predicted scores. TYPE: `ndarray`
`response`	Observed response values. TYPE: `ndarray`
`ax`	Axes to plot on. TYPE: `Axes \| None` DEFAULT: `None`
`point_size`	Marker size (matplotlib `s` parameter). TYPE: `float` DEFAULT: `0.5`
`alpha`	Marker transparency. TYPE: `float` DEFAULT: `1.0`
`title`	Optional title. TYPE: `str \| None` DEFAULT: `'Regression prediction'`

RETURNS	DESCRIPTION
`Axes`

plot_regression_prediction_binary ¶

plot_regression_prediction_binary(pred: ndarray, response: ndarray, *, ax: Axes | None = None, title: str | None = 'Regression prediction') -> matplotlib.axes.Axes

Plot 1-ECDF of predictions stratified by binary response class.

Mirrors the R plot_regression_prediction_binary which shows the inverted empirical CDF (1 - ECDF) for class 0 (blue) and class 1 (red), plus a KS D statistic annotation.

PARAMETER	DESCRIPTION
`pred`	Predicted scores. TYPE: `ndarray`
`response`	Binary response (0/1). TYPE: `ndarray`
`ax`	Axes to plot on. TYPE: `Axes \| None` DEFAULT: `None`
`title`	Optional title. TYPE: `str \| None` DEFAULT: `'Regression prediction'`

RETURNS	DESCRIPTION
`Axes`

plot_regression_qc ¶

plot_regression_qc(result: RegressionResult, response: ndarray | None = None, *, title: str | None = None, point_size: float = 0.5, alpha: float = 0.5) -> matplotlib.figure.Figure

Multi-panel QC figure for a single-motif regression result.

Creates a figure with three panels:

PSSM sequence logo
Spatial model
Prediction vs response scatter (continuous) or 1-ECDF (binary)

Mirrors plot_regression_qc from R.

PARAMETER	DESCRIPTION
`result`	Output of :func:`pyprego.regress_pwm`. TYPE: `RegressionResult`
`response`	Response variable. Required -- used for the prediction panel. TYPE: `ndarray \| None` DEFAULT: `None`
`title`	Overall figure title. Defaults to showing the consensus. TYPE: `str \| None` DEFAULT: `None`
`point_size`	Point size for scatter plot. TYPE: `float` DEFAULT: `0.5`
`alpha`	Transparency for scatter plot. TYPE: `float` DEFAULT: `0.5`

RETURNS	DESCRIPTION
`Figure`	The figure containing all panels.

RAISES	DESCRIPTION
`ValueError`	If response is `None`.

plot_regression_qc_multi ¶

plot_regression_qc_multi(result, response: ndarray | None = None, *, title: str | None = None, point_size: float = 0.01, alpha: float = 0.5) -> matplotlib.figure.Figure

Multi-panel QC figure for a multi-motif regression result.

For each motif, shows: PSSM logo, spatial model, and prediction panel. Also shows a score summary panel at the bottom.

Mirrors plot_regression_qc_multi from R.

PARAMETER	DESCRIPTION
`result`	Output of :func:`pyprego.regress_multiple_motifs`. TYPE: `MultiRegressionResult`
`response`	Response variable. Required -- used for prediction panels. TYPE: `ndarray \| None` DEFAULT: `None`
`title`	Overall figure title. TYPE: `str \| None` DEFAULT: `None`
`point_size`	Point size for scatter plots. TYPE: `float` DEFAULT: `0.01`
`alpha`	Transparency for scatter plots. TYPE: `float` DEFAULT: `0.5`

RETURNS	DESCRIPTION
`Figure`	The figure containing all panels.

RAISES	DESCRIPTION
`ValueError`	If response is `None` or result has no `models` attribute.