Dataset Viewer
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code: StreamingRowsError
Exception: ArrowInvalid
Message: Mismatching child array lengths
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
return get_rows(
dataset=dataset,
...<4 lines>...
column_names=column_names,
)
File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
return func(*args, **kwargs)
File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
yield from ds.decode(False) if ds.features else ds
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
for key, pa_table in self.ex_iterable._iter_arrow():
~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
for key, pa_table in self.generate_tables_fn(**gen_kwags):
~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 83, in _generate_tables
pa_table = _recursive_load_arrays(h5, self.info.features, start, end)
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 269, in _recursive_load_arrays
arr = _recursive_load_arrays(dset, features[path], start, end)
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/hdf5/hdf5.py", line 290, in _recursive_load_arrays
sarr = pa.StructArray.from_arrays(values, names=keys)
File "pyarrow/array.pxi", line 4304, in pyarrow.lib.StructArray.from_arrays
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: Mismatching child array lengthsNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
BulkGPT Benchmark Data
Benchmark datasets for the BulkGPT multi-foundation-model deconvolution benchmark: 42 datasets (30 pseudo-bulk + 12 real-bulk) in a unified H5 format. Companion code: RainyEricYe/bulkgpt-deconvolution-benchmark.
Contents
1_pseudo_bulk/ — 30 synthetic pseudo-bulk datasets (30 files)
Generated by aggregating single-cell RNA-seq references into pseudo-bulk mixtures with Dirichlet-sampled proportions and known ground truth.
| Subdirectory | # Datasets | Source |
|---|---|---|
cellxgene/ |
10 | CELLxGENE tissues (Liver, Heart, Brain regions, etc.) |
tabula_sapiens/ |
20 | Tabula Sapiens consortium organs |
2_real_bulk/ — 12 real-bulk validation datasets (24 files)
Real bulk RNA-seq cohorts with ground-truth cell-type proportions
(12 H5 + 12 *_gt.csv).
| Dataset | Notes |
|---|---|
sdy67 |
Purified PBMC mixtures (ImmPort SDY67) |
sweetwater |
PBMC, FACS-sorted ground truth |
huuki_myers |
DLPFC brain tissue |
demixsc_retina |
Retina |
altman_Arunachalam, altman_Hao, altman_TabulaSapiens |
PBMC cohorts (Altman et al.) |
finotello_Hao, hoek_Hao, hoek_purified_Hao, linsley_purified_Hao, morandini_Hao |
Hao et al. cohorts |
H5 Canonical Format
All H5 files follow a single convention (row-major):
bulk/values: (n_samples, n_genes) — bulk expression
bulk/rownames: genes
bulk/colnames: sample IDs
singleCellExpr/values: (n_cells, n_genes) — scRNA-seq reference
singleCellExpr/rownames: genes
singleCellExpr/colnames: cell barcodes
singleCellLabels/values: (n_cells,) — cell type per cell
ground_truth/values: (n_types, n_samples) — real-bulk GT (2_real_bulk)
bulkRatio/values: (n_samples, n_types) — pseudo-bulk GT (1_pseudo_bulk)
R note: R's
h5readtransposes 2-D arrays. Methods usingDeconUtils::getArgshandle this automatically; seefixed_scripts/in the code repo for patches.
Usage
pip install huggingface_hub
# Download all data (~10 GB)
huggingface-cli download yeruihku/bulkgpt-data --repo-type dataset --local-dir data
# Or via the code repo's helper script (also supports split selection)
git clone https://github.com/RainyEricYe/bulkgpt-deconvolution-benchmark.git
cd bulkgpt-deconvolution-benchmark
bash data/download_data.sh # all
bash data/download_data.sh real # real-bulk only
bash data/download_data.sh pseudo # pseudo-bulk only
Load in Python:
from core.data_loader import load_data # from the code repo
bundle = load_data("data/2_real_bulk/sdy67.h5")
bundle.bulk # (250 samples × 17,387 genes)
bundle.sc_ref # single-cell reference (AnnData)
bundle.gt # ground-truth proportions
Citation
@software{bulkgpt2026,
author = {Ye, Rui},
title = {Multi-Foundation Model Benchmark for Bulk RNA-seq Deconvolution},
year = {2026},
url = {https://github.com/RainyEricYe/bulkgpt-deconvolution-benchmark}
}
License
CC BY 4.0. Source datasets are public (CELLxGENE, Tabula Sapiens, GEO, ImmPort).
- Downloads last month
- 103