CiteSeqDataset #40
CiteSeqDataset #40
Conversation
…or "pbmc" and info about ADT counts (proteins markers), added as attributes of the dataset. This is the data from epitopes useful for having further labelling information.
|
I haven't had the time to create the preprocessed files for unit tests, but:
|
tests/test_scvi.py
Outdated
| @@ -30,7 +30,7 @@ def test_retina(): | |||
|
|
|||
|
|
|||
| def test_cbmc(): | |||
| run_benchmarks("cbmc", n_epochs=1, show_batch_mixing=False, save_path='tests/data/') | |||
| run_benchmarks("cite_seq_cbmc", n_epochs=1, show_batch_mixing=False, save_path='tests/data/') | |||
There was a problem hiding this comment.
Is cite_seq_cbmc the same data as cbmc once it's loaded? Can we just continue to call it cbmc then?
…or "pbmc" and info about ADT counts (proteins markers), added as attributes of the dataset. This is the data from epitopes useful for having further labelling information.
- Datasets might have multiple urls from which to download (ex. cite-Seq data): we might either specify `url`, `download_name` attributes or `urls`, `download_names`. Move check if file exists in `_download`.
- `CbmcDataset()` -> CiteSeqDataset('cbmc'), with information about epitopes.
- `PbmcDataset()` can be obtained with:
```
gene_dataset = concat_datasets(
Dataset10X("pbmc8k", save_path=save_path),
Dataset10X("pbmc4k", save_path=save_path)
)
```
So I removed data/PBMC
- From citeSeq methods there are actually 3 available datasets (cmbc, pbmc, and cd8). Since there are also 10X pbmc datasets, the `pbmc` nameis misleading in the `load_datasets` function. For now we leave as default romain's initial pbmc dataset, which consists in the concatenation of `pbmc8k` and `pbmc4k`
- `concat_datasets` test
So I removed data/PBMC
|
Codecov Report
@@ Coverage Diff @@
## master #40 +/- ##
==========================================
+ Coverage 89.08% 91.42% +2.33%
==========================================
Files 33 32 -1
Lines 1393 1388 -5
==========================================
+ Hits 1241 1269 +28
+ Misses 152 119 -33
Continue to review full report at Codecov.
|
|
closes #47 |
|
Very nice! |
I suggest this to complete CbmcDataset information.
This is the data from epitopes useful for having further labelling information.