Part 1: Breast cancer remains the leading cause of cancer mortality in women, largely due tometastatic recurrence. Recurrent tumors arise from residual tumor cells, which can survive cancer therapy and stay dormant for an extended period of time. Despite their clinical importance, the biology of tumor dormancy and recurrence are poorly understood. Using a HER2 inducible mouse model of breast cancer, we performed a CRISPR screen to identify novel epigenetic regulators of tumor dormancy and recurrence. We identified HDAC10 and KMT2E as recurrencepromoting genes; tumor cells with Hdac10 or Kmt2e loss were depleted during dormancy and their loss delayed recurrence. Conversely, we identified KDM1A as a recurrence-suppressing gene; cells with Kdm1a loss were enriched during dormancy and Kdm1a loss accelerated recurrence. Intriguingly, we discovered paradoxical functions of KAT7; cells with Kat7 loss were enriched during dormancy and recurrence, whereas Kat7 loss delayed recurrence, suggesting a potential biphasic effect on dormancy and recurrence. Additionally, genomic data for human breast cancer and genetically engineered mouse models of breast cancer were consistent with the findings for each gene. Collectively, we provide the first evidence to implicate HDAC10, KMT2E, KDM1A, and KAT7 in breast cancer dormancy and recurrence. Part 2: Successful CRISPR screens require an accurate understanding of background levels of sgRNA skewing in an experimental system of interest, yet this type of characterization has not been performed for in vivo CRISPR screens in mouse models of cancer. To address this gap, we performed an in vivo CRISPR screen with a negative control sgRNA library that was composed of non-targeting and non-functional sgRNAs. This revealed a progressive skewing of a negative sgRNA distribution throughout the time course, as well as an increase in intratumoral heterogeneity. Based on these results, we developed a novel pair-wise comparison analytical method that performed comparably to the-state-of-the-art method, MAGeCK. Additionally, we developed novel scoring methods for enrichment-based screen data analysis to detect clonal enrichment. Together, we provide in vivo negative library screen data, which will be a valuable reference for understanding the null distribution of sgRNAs in vivo screen, and provide new analytical methods and metrics as tools for the analysis of in vivo CRISPR screens