sBlot

sBlot Documentation

Introduction

This document explains how to use the sBlot package to visualise the results of an sBayes analysis.

sBayes is a clustering method that identifies groups with similar features while accounting for similarities due to known confounders.

Static visualisations

Users can create five main types of static plots:

1. Weight plots visualise the weights assigned to each confounder and each of the clusters.
2. Preference plots show the distribution of a feature in each cluster and in each group within each confounder.
3. Pie plots show the membership of objects to clusters.
4. LOO plots compare the Pareto-Smoothed Importance Sampling Leave-One-Out Cross-Validation (PSIS-LOO) for models with different numbers of clusters.
5. Maps visualise the spatial allocation of objects to clusters.

Users define all plotting parameters in two configuration files:

• config_plot.yaml specifies the input data and results of the sBayes analysis to visualise, the plots to generate, and the analytical tasks to perform for each plot.
• config_style.yaml defines the graphical style, including colors and font sizes.

Typically, users provide all parameters in the config_plot.yaml file, but use the pre-set style parameters for config_style.yaml, modifying only selected parameters as needed.

Interactive explorer

Alternatively, users can also explore the posterior distribution in an interactive explorer.

Documentation

Installation	How to install sBlot
Quick start	Creating plots in a few steps
Plot types	Overview of all available plot types
Plot configuration	Full config_plot.yaml reference
Style configuration	Full config_style.yaml reference
Interactive explorer	Browser-based posterior exploration

This site is open source. Improve this page.