---
title: IDEAS
canonical_url: "https://www.modelscope.cn/datasets/User040218/IDEAS"
md_url: "https://www.modelscope.cn/datasets/User040218/IDEAS.md"
repository: User040218/IDEAS
last_updated: 2026-11-01
license: other
storage_size: "3.5 GB"
downloads: 426
stars: 0
---

# IDEAS

> IDEAS - User040218 在 ModelScope 开源的数据集。IDEAS: 84 ENCODE signal tracks over hg19 200 bp windows with window sequences (84-target regression).

User040218/IDEAS 是 ModelScope 魔搭社区上的数据集，存储大小 3.5 GB，采用 other 许可。

- **Repository**: User040218/IDEAS
- **License**: other
- **Storage size**: 3.5 GB
- **Downloads**: 426
- **Stars**: 0
- **Last updated**: 2026-11-01

Source: https://www.modelscope.cn/datasets/User040218/IDEAS

---

# IDEAS

**84 ENCODE signal tracks over hg19 200 bp windows with window sequences** — regression.

IDEAS jointly characterizes epigenetic dynamics across multiple human cell types. This release uses the 84 ENCODE uniformly processed signal tracks listed in the paper's Supplementary Table S1 (hg19, `*.norm5.rawsignal.bw`).

Each track takes its maximum value within a 200 bp window; together with the window's DNA sequence this forms an 84-target regression task over 13,774,417 windows in a single train split. Windows missing all tracks are removed, and a single missing track is set to 0 before the log2(x + 1) transform.

Links: [IDEAS (GitHub)](https://github.com/yuzhang123/IDEAS).

## Splits

Subset: `default`

| Split | Rows |
|---|---:|
| train | 13,774,417 |

## Format

Parquet. Main columns:

| Column | Type | Description |
|---|---|---|
| `sequence` | string | 200 bp DNA sequence (hg19) |
| `labels` | list<float32>[84] | 84 signal values (log2(x + 1)) |
| `chrom` / `start` | string / int64 | Window coordinates |

## Usage

Example with [DNALLM](https://github.com/zhangtaolab/DNALLM): subset `default`, regression, `num_labels=84`.

```python
from modelscope.msdatasets import MsDataset
from dnallm import DNADataset

ds = MsDataset.load("User040218/IDEAS", subset_name="default")
# tokenizer: the tokenizer of your DNALLM model; set max_length to match its tokenization
data = DNADataset(ds, tokenizer=tokenizer, max_length=202)
```

## Notes

- The window count differs from the paper's 13,763,197 and the paper's exact exclusion rule has not been confirmed; all 15,181,614 raw windows are kept in `windows/`.
- Only a train split is provided.

## Source

If you use this dataset, please cite:

- Zhang Y, An L, Yue F, Hardison RC. Jointly characterizing epigenetic dynamics across multiple human cell types. Nucleic Acids Research 2016;44(14):6721-6731. doi:10.1093/nar/gkw278

## License

Declared as `other`: use of the data follows the terms of the original data providers listed in Source.