Cancer starts years to decades before diagnosis. Yet, early stages of tumor development are poorly understood. In particular, the precancer landscape of chromosome instability (CIN) is understudied, due to the scarcity of precursor lesions and methods to analyze CIN in single cells. The research presented here will address how the instability rate of the genome, followed by selection, determines whether a CIN-driven tumor arises. I hypothesize that the onset of CIN is the tipping point that governs the fate of a precancer clone. Using spatially aware and single-cell multiome methods that we developed, we will dissect essential factors that lead to malignancy via CIN, including extrachromosome circular DNAs. We will analyze two solid tissues at opposite ends of the cell turnover spectrum, which deeply diverge in regeneration capacity, clonality and constraints, namely the brain and the gut. Aim 1 will assess the instability rate of the human genome using the postmortem brain as a model tissue. CIN patterns vary greatly in frequency and type across brain tumors; we will compare the background incidence of fundamental types of CIN in normal cells to CIN signatures after selection has acted, in diagnosed tumors of the same lineages. Aim 2 will dissect CIN in space and time from the cell of origin, to identify what discriminates nascent clones that may cause cancer from those that do not. Timecourses in mouse models will reveal how CIN contributes to selective advantages and how nascent CIN-positive clones escape elimination by immune cells. Aim 3 leverages human precursor lesions to probe the role of CIN as a central gatekeeper between tumor initiation and growth, using the gut as a longitudinal in vivo system to directly capture the tipping point to cancer. Unstable Genome will provide the first systematic quantification of the early development of chromosomally unstable lesions and reveal the principles that govern CIN-driven clone expansion in vivo.