File size: 2,490 Bytes
12ce8a1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
---
license: mit
task_categories:
  - other
tags:
  - biology
  - gene-regulatory-networks
  - single-cell
  - glioblastoma
  - perturbation
  - grn-inference
size_categories:
  - 10K<n<100K
---

# scRegNet Data & Results

Processed data and results for **scRegNet**: perturbation-calibrated gene representations
for context-specific gene regulatory network inference in glioblastoma.

**Code**: [github.com/mmTheBest/scRegNet](https://github.com/mmTheBest/scRegNet)

## Contents

### Processed Data (`data/processed/`)

| File | Description | Size |
|------|-------------|------|
| `gene_universe.json` | G\* gene universe (18,716 genes) | 252 KB |
| `hgfb_raw.npz` | H-GFB features (18,716 × 1,152) | 83 MB |
| `microglia_labels.npz` | Microglia perturbation labels (38 regulators) | 800 KB |
| `k562_labels.npz` | K562 perturbation labels (9,611 regulators) | 19 MB |
| `embeddings/*.npz` | 6 embedding sources (18,716 × 256 each) | 116 MB |
| `atlas/celltype_means.npz` | Cell-type mean expression (11 types) | 784 KB |
| `atlas/neftel_subtypes.npz` | Neftel subtype means (6 subtypes) | 3.2 MB |
| `validation/*.json` | ChIP-Atlas, DoRothEA, TRRUST, canonical TFs | 9 MB |
| `validation/depmap_ceres.npz` | DepMap CERES scores (50 GBM lines) | — |
| `validation/msigdb_genesets.json` | MSigDB gene sets (7,540 sets) | — |

### Results (`results/`)

| Directory | Description |
|-----------|-------------|
| `stage_b/` | Stage B model outputs (6 embeddings + 3 controls) |
| `stage_c/` | Stage C GRN predictions (6 sources × 17 contexts, parquet) |
| `competing/` | 8 competing method outputs |
| `experiments/exp_1-9/` | All 9 experiment results (JSON) |
| `figures/` | 7 main + 6 extended figures (PDF + PNG) |

### Not Included (Regeneratable)

- Raw count matrices (`data/processed/stage_c/`, 4.3 GB) — regenerate from GEO GSE274546
- Per-cell expression matrices (`data/processed/atlas/celltype_expr/`, 1.8 GB) — regenerate via `05a_prepare_atlas.py`

## Usage

```python
from huggingface_hub import snapshot_download

# Download everything
snapshot_download("mm1109/scRegNet", repo_type="dataset", local_dir="./scRegNet-data")

# Load embeddings
import numpy as np
data = np.load("scRegNet-data/data/processed/embeddings/geneformer.npz")
embeddings = data["embeddings"]  # (18716, 256)
genes = data["genes"]            # (18716,)

# Load GRN predictions
import pandas as pd
grn = pd.read_parquet("scRegNet-data/results/stage_c/geneformer.parquet")
```

## License

MIT