Saccharomyces cerevisiae is a model organism used for studying fundamental eukaryotic processes and genome function. While essentiality studies provide important insight into gene function, biological pathways, and evolution, they often focus on entire genes, thereby ignoring variation in essentiality within and beyond gene boundaries. SAturated Transposon Analysis in Yeast (SATAY) is a Transposon Insertion Sequencing (TIS) technique that can measure essentiality across the entire genome of S. cerevisiae. However, the resulting data is noisy and sparse, and is affected by strong insertion biases. Most existing methods for analyzing TIS data either rely on known gene annotations or are optimized for bacterial datasets. Methods optimized for bacterial datasets are difficult to apply to SATAY data due to its higher sparsity and distinct insertion patterns. We therefore aim to develop a method tailored to SATAY data that does not rely on predefined annotations. In this study, we developed a change-point detection (CPD) algorithm to identify genomic regions with distinct levels of essentiality directly from SATAY data. We explored whether an autoencoder could improve the quality of SATAY data by denoising and imputing missing values. Although the autoencoder appeared to denoise the data, it did not perform meaningful imputation. Moreover, CPD applied to the AE output was outperformed by a CPD algorithm that modeled raw SATAY data as a zero-inflated negative binomial (ZINB) distribution. The ZINB-based CPD algorithm achieved better results by explicitly accounting for sparsity, overdispersion, and insertion biases. The regions produced by our CPD algorithm align with biological expectations, though they are oversegmented and remain affected by known biases in the data. Despite these limitations, our results show that CPD applied to SATAY data is a promising first step towards identifying essential genomic regions beyond predefined gene boundaries across the whole genome. ... Read more