T cell models¶
Overview¶
The T Cell model is a building model for 3D cells with protein patterns, and can be used to develop 4D movies. The model is based on the T cell model in Royal et al. 2016. We assume different cells have similar cell shape and can be mapped to a template. This lends the model to be useful for quantitative analysis of proteins in T cells, as well as other cells. Similar to other models in CellOrganizer, there are two parts: training and synthesis. In training, a morphing model is trained from the original images. In synthesis, images of cells are synthesized from the trained model and include the protein pattern.
We have a protocol chapter that describes how the images are generated, annotated and analysized:
- Ambler, R., Ruan, X., Murphy, R. F. and Wülfing, C., 2017. Systems Imaging of the Immune Synapse. The Immune Synapse: Methods and Protocols, pp.409-421.
The usage of the model is part of the chapter (section 3.7 & 3.8)
Training¶
The training part requries T cell movies and the annotation of the synapse positions of the T cells as input. It can be further broken down into the following steps: cropping, segmentation, rigid alignment, non-rigid alignment (morphing) and model-building.
The training demo included, demo3Dtcell_train, trains a protein distribution model following the approach described in
- Roybal, T. E. Buck, X. Ruan, B. H. Cho, D. J. Clark, R. Ambler, H. M. Tunbridge, J. Zhang, P. Verkade, C. Wülfing, and R. F. Murphy (2016) Computational spatiotemporal analysis identifies WAVE2 and Cofilin as joint regulators of costimulation-mediated T cell actin dynamics. Science Signaling 9:rs3. doi: 10.1126/scisignal.aad4149.
The slowest step, which takes approximately 1 min per cell per frame, is the alignment of each cell to the standardized template. This demo uses 46 cells, so it will run for about an 1 hour on a single core.
Synthesis¶
In the synthesis part, a T cell model is required as input. Should there be a specified shape to the cells, then a cell shape model is also required. The synthesis can be further broken down into the following steps: voxel sampling, shape registration, and voxel mapping.
The synthesis demo included, demo3Dtcell_synth, sythesizes from a T cell model. The demo takes in two models, one model containing the cell and nuclear shape models, and the other containing a T cell protein shape model.
Annotation File¶
Each annotation file contains the information of chosen cell pairs for a movie (each frame is a 3D stack). An example annotation file is shown below. The annotation file should be formated as described below:
- Each row represents a cell in a specific time point. And the adjacent rows without blank row separation represents a cell in different time points. Different cells are separated with blank row(s).
- In column 1, the name of the image file (depending on how the images are acquired, there may be one file per time point or multiple time points in a single file).
- In column 2, the number of the channel within that file that contains the GFP fluorescence for that time point (if each time point is in a separate file, this would typically be channel 1; if multiple time points are in the same file, this would typically be the frame number).
- In columns 3–7, the X coordinate of the left end point, the Y coordinate of the left end point, the X coordinate of the right end point, and the Y coordinate of the right end point for the synapse in that time point.
- In column 8, the time difference for that frame relative to time point 0.
