Monocle 2 is a computational method that uses reversed graph embedding (RGE) to accurately reconstruct complex single-cell trajectories without prior knowledge of cell fates or branch points. It was tested on two studies of blood development and revealed that mutations in key lineage transcription factors can divert cells to alternative fates. Monocle 2 outperforms previous versions and other methods in reconstructing accurate and robust trajectories. It identifies branch points automatically and does not require user input for the number of branches or cell fates. Monocle 2 uses an unsupervised procedure called "dpFeature" to identify genes that define biological processes. It then applies RGE to learn a principal graph that describes the trajectory. This method allows for the reconstruction of complex trajectories with multiple branches. When applied to myoblasts, Monocle 2 revealed a trajectory with a single branch point leading to two outcomes, with different gene expression patterns on each branch. A global analysis of branch-dependent gene expression confirmed that cells along these two outcomes differed in the expression of 887 genes. Monocle 2 was also tested on a dataset of blood development and showed that it can accurately reconstruct trajectories with multiple branches. It outperformed other methods in terms of accuracy and robustness. Monocle 2 can also be used for dimensionality reduction and graph learning, with alternative algorithms such as SimplePPT, SGL-tree, and DDRTree. These algorithms were tested on various datasets and showed high concordance in trajectory reconstruction. Monocle 2 was further validated by analyzing genetic perturbations in mice lacking key transcription factors. These perturbations caused cells to be diverted to alternative fates, demonstrating the method's ability to detect such changes. Monocle 2 was also compared to other methods such as Wishbone, DPT, and SLICER, and showed superior performance in terms of accuracy and robustness. It was able to correctly assign cells to branches and produce consistent results even when using subsampled data. Monocle 2's ability to reconstruct complex trajectories with multiple branches makes it a powerful tool for studying cell fate decisions. It can be applied to various types of single-cell data, including RNA-seq, chromatin accessibility, and 3D structure analysis. The method's robustness and accuracy make it a valuable tool for understanding the regulatory mechanisms that govern cell differentiation and development.Monocle 2 is a computational method that uses reversed graph embedding (RGE) to accurately reconstruct complex single-cell trajectories without prior knowledge of cell fates or branch points. It was tested on two studies of blood development and revealed that mutations in key lineage transcription factors can divert cells to alternative fates. Monocle 2 outperforms previous versions and other methods in reconstructing accurate and robust trajectories. It identifies branch points automatically and does not require user input for the number of branches or cell fates. Monocle 2 uses an unsupervised procedure called "dpFeature" to identify genes that define biological processes. It then applies RGE to learn a principal graph that describes the trajectory. This method allows for the reconstruction of complex trajectories with multiple branches. When applied to myoblasts, Monocle 2 revealed a trajectory with a single branch point leading to two outcomes, with different gene expression patterns on each branch. A global analysis of branch-dependent gene expression confirmed that cells along these two outcomes differed in the expression of 887 genes. Monocle 2 was also tested on a dataset of blood development and showed that it can accurately reconstruct trajectories with multiple branches. It outperformed other methods in terms of accuracy and robustness. Monocle 2 can also be used for dimensionality reduction and graph learning, with alternative algorithms such as SimplePPT, SGL-tree, and DDRTree. These algorithms were tested on various datasets and showed high concordance in trajectory reconstruction. Monocle 2 was further validated by analyzing genetic perturbations in mice lacking key transcription factors. These perturbations caused cells to be diverted to alternative fates, demonstrating the method's ability to detect such changes. Monocle 2 was also compared to other methods such as Wishbone, DPT, and SLICER, and showed superior performance in terms of accuracy and robustness. It was able to correctly assign cells to branches and produce consistent results even when using subsampled data. Monocle 2's ability to reconstruct complex trajectories with multiple branches makes it a powerful tool for studying cell fate decisions. It can be applied to various types of single-cell data, including RNA-seq, chromatin accessibility, and 3D structure analysis. The method's robustness and accuracy make it a valuable tool for understanding the regulatory mechanisms that govern cell differentiation and development.