Love-thy-neighbor: Neural networks for tracking and lineage tracing in budding yeast

Abstract

Tracking and lineage tracing are widely needed tasks in biological image analysis. For cells that grow and divide, tracking is challenging because cells change in number, shape, and size throughout a recording. As the time interval between images increases, it becomes more difficult to establish correspondences between cells across timepoints. Consequently, tracking has to be performed between consecutive or temporally close images, which leads to exponentially decreasing tracking accuracy and thus high sensitivity to error rates. For budding yeast, this challenge is further heightened by the similarity of cells in colonies, their dense packing, the asymmetric nature of cell divisions, and movement due to growth of the colony. A related task, lineage tracing, is similarly challenging without fluorescent markers due to multiple potential mother cells surrounding a new daughter cell. Here, we present neural networks for budding yeast tracking and lineage tracing, named LYN-track and LYN-trace, respectively. These methods leverage fine geometric features of cells and their neighborhoods. To train and test the algorithms, we recorded and annotated new budding and fission yeast microscopy movies (78,852 frame-to-frame tracklets, 2,512 images), which we make freely available. On these and existing datasets, our neural network-based methods demonstrate robust, above state-of-the-art performance. Both tools have been integrated into graphical user interfaces (GUIs), available on Github, and can be straightforwardly retrained with custom data if desired.