Leveraging Lineage Barcodes as Natural Augmentations for Contrastive Learning of Cell Fate in scRNA-seq Data

Deciphering how cells commit to future fates is essential for developing precision therapeutics that can reprogram stem cells or modulate immune functions. However, isolating these fate-determining signals in single-cell lineage tracing (scLT) remains challenging because differentiation programs are often confounded by unrelated processes like the cell cycle. To address this, we introduce Lineage-aware Contrastive Learning (LCL), a framework that treats inheritable lineage barcodes as a "natural" data augmentation to isolate subtle, lineage-specific signals. LCL utilizes a semi-supervised architecture to align unlabeled cells, facilitating the transfer of lineage structures to clinical datasets where explicit barcoding is unavailable. We demonstrate LCL’s utility by predicting future cell-type compositions from early-time points, effectively modeling longitudinal fate commitment from cross-sectional data. Benchmarking on hematopoietic and fibroblast systems shows that LCL significantly outperforms standard models like scVI, establishing contrastive learning as a scalable paradigm for understanding and potentially manipulating cellular differentiation.

Understanding how stem cells choose their future identities is essential for developing precise medical treatments that can repair tissues or boost the immune system. However, pinpointing the exact genetic signals that dictate a cell’s destination is incredibly difficult because other daily cell activities, like dividing and growing, create overwhelming background noise. While scientists can insert microscopic "barcodes" into cells to track family trees over time, this special tracking is missing from the vast majority of standard medical datasets. To bridge this gap, we created Lineage-aware Contrastive Learning (LCL), an artificial intelligence framework that acts like a specialized information filter. LCL treats inherited barcodes as a natural guide, teaching the computational method to ignore the background noise and hunt specifically for the faint genetic signatures shared by related cells. Crucially, our method can then transfer this learned understanding to standard biomedical data where tracking labels are unavailable, helping researchers accurately predict a cell's ultimate fate from its earliest stages. This new tool outperforms existing methods, offering a clearer map for guiding and manipulating stem cells to treat human diseases.