Topyfic

- due to changes in `random.shuffle()`, any version >= 0.4.17 only compatible with python >= 3.9

- Add M-A plot to compare two topics based on their gene weights and calculate the modified z-score
- Compare two topics based on their gene weights
- Embedded other models to be able to use downstream analysis
- Add different statistical comparisons to compare topics i.e. Jensen–Shannon divergence
- Save train and topModel objects as a h5 file to be able to read and work it without using Python
- update topic_trait_relationships_heatmap()

- Adding more options to tune the parameters you used to learn LDA models
- Adding options to combine multiple trains which help you to save time

Topyfic is a Python package designed to identify reproducible latent Dirichlet allocation (LDA) using Leiden clustering and harmony for single-cell epigenomics data

Topyfic has a lot of options for downstream analysis
- GSEA
- GO/KEGG
- REACTOME
- Structure plots
- Pie chart
- Comparison across topics within/across multiple runs
- and lots of other visualization options

Topyfic is a Python package designed to identify reproducible latent Dirichlet allocation (LDA) using Leiden clustering and harmony for single-cell epigenomics data