AutoZI is a deep generative model adapted from scVI allowing a gene-specific treatment of zero-inflation. Plots relating to batch effects and gene expression levels are also generated. Ideally, data should be provided in .h5ad format, which is native to the Anndata python package. Alternatively, .h5, .csv or.mtx files can be provided if their formatting is suitable.

Example use case: In single-cell RNA sequencing data, biological processes or technical factors may induce an overabundance of zero measurements. Existing probabilistic approaches to interpreting these data either model all genes as zero-inflated, or none. But the overabundance of zeros might be gene-specific. AutoZI can distinguish between zero-inflated and non-zero inflated genes.