Abstract 2487: LINMAP: A framework for comprehensive cell division lineage reconstruction via molecular barcoding and machine learning

Abstract Purpose: Reconstructing cell division lineages across generations is central to understanding developmental biology, cancer progression, cell differentiation, drug resistance, and tissue aging. By integrating CRISPR-Cas9 induced gene editing with single cell RNA-Seq, high-resolution lineage tracing through molecular barcoding becomes feasible. However, tracking mutation evolution across successive guide RNA edits, especially with homing guide RNA (hgRNA) technology, poses challenges due to ongoing dynamic retargeting and genetic modifications that complicate lineage reconstruction. Methods: To overcome these challenges, we developed LINMAP, a computational framework that applies machine learning to trace and map evolving mutations across cell divisions, reconstructing intermediate histories. LINMAP calculates mutation transition probabilities at each division to identify potential paths for mutation accumulation and defines cell-cell distances based on these transition probabilities, enabling precise division count measurements between cell pairs. Results: We created an MMTV-PyMT (mouse mammary tumor virus-polyoma middle tumor-antigen)-Cas9 tracing cell line with dual hgRNAs to investigate mutation evolution networks. LINMAP accurately inferred mutation evolution and predicted allele frequencies, aligning well with both in vitro data from MMTV-PyMT cell lines and computational simulations. Moreover, in a separate in silico experiment simulating lineage history from a progenitor, LINMAP reconstructed the cell division lineage with high precision. Applying LINMAP to a CRISPR-Cas9 lineage-traced lung cancer mouse model, we successfully traced tumors of varying clonality within a single mouse. Based on lineage characteristics, LINMAP differentiated tumors, reconstructed their complete lineage, and identified each tumor’s monoclonal or polyclonal origins. It recreated a timeline of tumor emergence and tracked cell state transitions, capturing cell state shifts during cancer progression. These findings aligned with gene expression-based cell state trajectories from the original study, further validating LINMAP's accuracy. Conclusions: LINMAP integrates mutation ratios and transition probabilities at each cell division, surpassing precision than existing methods. It enables comprehensive single cell lineage tracing, providing insights into tumor evolution and complexity, tracking tumor development, distinguishing metastatic sites from primary tumors, and studying drug resistance through subclonal selection over time. Citation Format: Lin Wang, Zheng Yin, Wenjuan Dong, Jianting Sheng, Li Yang, Kun Han, Dingcheng Gao, Stephen T. C. Wong. LINMAP: A framework for comprehensive cell division lineage reconstruction via molecular barcoding and machine learning [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2487.