Question: How can I customize the parameters for t-Distributed Stochastic Neighbor Embedding (t-SNE) in Cell Ranger, and what are the default values?
Answer: You can modify the following parameters related to t-SNE by passing a CSV file to cellranger reanalyze:
| Parameter | Type | Default Value | Recommended Range | Description |
| tsne_input_pcs | int | null | Cannot be set higher than the num_principal_compsparameter. | Subset to top N principal components for TSNE. Change this parameter if you want to see how the TSNE plot changes when using fewer PCs, independent of the clustering / differential expression. You may find that TSNE is faster and/or the output looks better when using fewer PCs. |
| tsne_perplexity | int | 30 | 30-50 | TSNE perplexity parameter (see [the TSNE FAQ](https://lvdmaaten.github.io/tsne/) for more details). When analyzing 100k+ cells, increasing this parameter may improve TSNE results, but the algorithm will be slower. |
| tsne_theta | float | 0.5 | Must be between 0 and 1. | TSNE theta parameter (see [the TSNE FAQ](https://lvdmaaten.github.io/tsne/) for more details). Higher values yield faster, more approximate results (and vice versa). The runtime and memory performance of TSNE will increase dramatically if you set this below 0.25. |
| tsne_max_dims | int | 2 | Must be 2 or 3. | Maximum number of TSNE output dimensions. Set this to 3 to produce both 2D and 3D TSNE projections (note: runtime will increase significantly). |
| tsne_max_iter | int | 1000 | 1000-10000 | Number of total TSNE iterations. Try increasing this if TSNE results do not look good on larger numbers of cells. Runtime increases linearly with number of iterations. |
| tsne_stop_lying_iter | int | 250 | Cannot be set higher than tsne_max_iter. | Iteration at which TSNE learning rate is reduced. Try increasing this if TSNE results do not look good on larger numbers of cells. |
| tsne_mom_switch_iter | int | 250 | Cannot be set higher than tsne_max_iter. | Iteration at which TSNE momentum is reduced. Try increasing this if TSNE results do not look good on larger numbers of cells. Cannot be set higher than tsne_max_iter. |
| random_seed | int | 0 | any integer | Random seed. Due to the randomized nature of the algorithms, changing this will produce slightly different results. If the TSNE results don't look good, try running multiple times with different seeds and pick the TSNE that looks best. |
For more information please see Customized Secondary Analysis using cellranger reanalyze .