Datasets:
Reformat headers
Browse files
README.md
CHANGED
|
@@ -21,23 +21,22 @@ Code to download and process this dataset is available in: https://github.com/se
|
|
| 21 |
|
| 22 |
Dataset structure is originally from [AnnData](https://anndata.readthedocs.io/en/latest/index.html),
|
| 23 |
|
|
|
|
| 24 |
|
| 25 |
-
##
|
| 26 |
-
|
| 27 |
-
### `bladder_smartseq2_expression.parquet`
|
| 28 |
|
| 29 |
`bladder_smartseq2_expression.parquet` is a 2,432 rows x 21,069 columns dataset. Each row is a single cell's gene expression across 21,069 mouse genes. This is typically the `X` matrix for ML modeling, and would need to be randomly split for test/train/validation sets.
|
| 30 |
|
| 31 |
|
| 32 |
|
| 33 |
-
|
| 34 |
|
| 35 |
`bladder_smartseq2_sample_metadata.parquet` is a 2,432 rows x 30 columns dataset. Each row represents the metadata for a single cell, e.g. what mouse it came from (`donor_id`), the sex of the mouse, number of genes expressed (`n_genes`), number of total read counts per cell (`n_counts`), cell type annotation (`cell_type`), age of the mouse (`age` or also `development_stage`)
|
| 36 |
|
| 37 |
|
| 38 |
|
| 39 |
|
| 40 |
-
|
| 41 |
|
| 42 |
`bladder_smartseq2_feature_metadata.parquet` is a 21,069 rows x 11 columns dataset. Each row represents the metadata for each gene, e.g. number of cells expressing it (`n_cells`), mean gene expression (`means`), if it's a highly variable gene (`highly_variable`), the type of the feature (`feature_type`)
|
| 43 |
|
|
@@ -45,7 +44,7 @@ Dataset structure is originally from [AnnData](https://anndata.readthedocs.io/en
|
|
| 45 |
|
| 46 |
|
| 47 |
|
| 48 |
-
|
| 49 |
|
| 50 |
`bladder_smartseq2_unstructured_metadata.json` is a key-value store of unstructured metadata information about the dataset.
|
| 51 |
|
|
@@ -53,7 +52,7 @@ Dataset structure is originally from [AnnData](https://anndata.readthedocs.io/en
|
|
| 53 |
|
| 54 |
|
| 55 |
|
| 56 |
-
|
| 57 |
|
| 58 |
`bladder_smartseq2_projection_*.parquet` are transformations of the expression data using either PCA (first 50 PCs), tSNE (2 dimensions for visualizationA), or UMAP (2 dimensions for visualization).
|
| 59 |
|
|
|
|
| 21 |
|
| 22 |
Dataset structure is originally from [AnnData](https://anndata.readthedocs.io/en/latest/index.html),
|
| 23 |
|
| 24 |
+
Descriptions of each data file is below.
|
| 25 |
|
| 26 |
+
## `bladder_smartseq2_expression.parquet`
|
|
|
|
|
|
|
| 27 |
|
| 28 |
`bladder_smartseq2_expression.parquet` is a 2,432 rows x 21,069 columns dataset. Each row is a single cell's gene expression across 21,069 mouse genes. This is typically the `X` matrix for ML modeling, and would need to be randomly split for test/train/validation sets.
|
| 29 |
|
| 30 |
|
| 31 |
|
| 32 |
+
## `bladder_smartseq2_sample_metadata.parquet`
|
| 33 |
|
| 34 |
`bladder_smartseq2_sample_metadata.parquet` is a 2,432 rows x 30 columns dataset. Each row represents the metadata for a single cell, e.g. what mouse it came from (`donor_id`), the sex of the mouse, number of genes expressed (`n_genes`), number of total read counts per cell (`n_counts`), cell type annotation (`cell_type`), age of the mouse (`age` or also `development_stage`)
|
| 35 |
|
| 36 |
|
| 37 |
|
| 38 |
|
| 39 |
+
## `bladder_smartseq2_feature_metadata.parquet`
|
| 40 |
|
| 41 |
`bladder_smartseq2_feature_metadata.parquet` is a 21,069 rows x 11 columns dataset. Each row represents the metadata for each gene, e.g. number of cells expressing it (`n_cells`), mean gene expression (`means`), if it's a highly variable gene (`highly_variable`), the type of the feature (`feature_type`)
|
| 42 |
|
|
|
|
| 44 |
|
| 45 |
|
| 46 |
|
| 47 |
+
## `bladder_smartseq2_unstructured_metadata.json`
|
| 48 |
|
| 49 |
`bladder_smartseq2_unstructured_metadata.json` is a key-value store of unstructured metadata information about the dataset.
|
| 50 |
|
|
|
|
| 52 |
|
| 53 |
|
| 54 |
|
| 55 |
+
## `bladder_smartseq2_projection_*.parquet`
|
| 56 |
|
| 57 |
`bladder_smartseq2_projection_*.parquet` are transformations of the expression data using either PCA (first 50 PCs), tSNE (2 dimensions for visualizationA), or UMAP (2 dimensions for visualization).
|
| 58 |
|