File size: 6,575 Bytes
d67f21a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
---
pretty_name: InterPro Entries
license: other
tags:
- biology
- protein
- protein-family
- protein-domain
- interpro
- ontology
- parquet
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*.parquet
  - split: test
    path: data/test-*.parquet
---

# InterPro Entries

This dataset contains a viewer-friendly Parquet table derived from the InterPro current release files in this repository. Each row is one InterPro entry from `current_release/interpro.xml.gz`.

The original source release files remain in the repository. The default `datasets` configuration uses the normalized Parquet files under `data/` so that the Hugging Face Dataset Viewer and `load_dataset()` can read the entries directly.

Large release artifacts such as `current_release/match_complete.xml.gz`, `current_release/protein2ipr.dat.gz`, and `current_release/sites.xml.gz` are preserved as source files but are not loaded by the default table.

## Splits

The split is deterministic by InterPro identifier: `sha256(interpro_id) % 10`. Bucket `0` is `test`; buckets `1` through `9` are `train`.

| Split | Rows |
|---|---:|
| train | 46,440 |
| test | 5,049 |
| total | 51,489 |

## Dataset Statistics

| Field | Value |
|---|---:|
| InterPro release | `108.0` |
| Release date | `29th January 2026` |
| Entries | 51,489 |
| Member database rows | 18 |
| Entries with GO mappings | 14,799 |
| InterPro-to-GO mapping rows | 30,200 |
| Entries with structures | 30,172 |
| Entries with publications | 38,194 |

| Entry type | Rows |
|---|---:|
| Family | 27,308 |
| Domain | 19,276 |
| Homologous_superfamily | 3,510 |
| Conserved_site | 768 |
| Repeat | 395 |
| Active_site | 133 |
| Binding_site | 82 |
| PTM | 17 |

| GO category | Mappings |
|---|---:|
| molecular_function | 13,802 |
| biological_process | 11,059 |
| cellular_component | 5,339 |

## Usage

Install the Hugging Face Datasets library:

```bash
pip install datasets
```

Load all splits:

```python
from datasets import load_dataset

ds = load_dataset("LiteFold/InterPro")
print(ds)

row = ds["train"][0]
print(row["interpro_id"], row["name"], row["entry_type"])
```

Load one split:

```python
from datasets import load_dataset

train = load_dataset("LiteFold/InterPro", split="train")
test = load_dataset("LiteFold/InterPro", split="test")
```

Stream rows without downloading the full table first:

```python
from datasets import load_dataset

stream = load_dataset("LiteFold/InterPro", split="train", streaming=True)
for row in stream.take(5):
    print(row["interpro_id"], row["short_name"], row["protein_count"])
```

Filter entries with GO mappings:

```python
from datasets import load_dataset

ds = load_dataset("LiteFold/InterPro", split="train")
with_go = ds.filter(lambda row: row["go_count"] > 0)
print(with_go[0]["interpro_id"], with_go[0]["go_ids"])
```

Filter protein domains with PDB structures:

```python
from datasets import load_dataset

ds = load_dataset("LiteFold/InterPro", split="train")
domains_with_structures = ds.filter(
    lambda row: row["entry_type"] == "Domain" and row["structure_count"] > 0
)
print(domains_with_structures[0]["interpro_id"], domains_with_structures[0]["pdb_ids"][:5])
```

Load release database metadata directly:

```python
import pandas as pd
from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="LiteFold/InterPro",
    repo_type="dataset",
    filename="metadata/database_info.parquet",
)
database_info = pd.read_parquet(path)
print(database_info)
```

## Columns

| Column | Description |
|---|---|
| `interpro_id` | InterPro accession, such as `IPR000001`. |
| `interpro_numeric_id` | Numeric portion of `interpro_id`. |
| `name` | Full InterPro entry name. |
| `short_name` | Short InterPro entry name from the XML attribute. |
| `entry_type` | Entry class, such as `Family`, `Domain`, or `Homologous_superfamily`. |
| `protein_count` | Number of proteins matched by the entry in the release XML. |
| `is_llm` | Whether the entry is marked as LLM-generated in the source XML. |
| `is_llm_reviewed` | Whether the LLM marker is reviewed in the source XML. |
| `abstract` | Normalized text from the entry abstract. |
| `go_ids` | GO identifiers mapped to the InterPro entry. |
| `go_terms` | GO term names corresponding to `go_ids`. |
| `go_categories` | GO namespaces corresponding to `go_ids`. |
| `go_count` | Number of mapped GO terms. |
| `member_databases` | Member databases contributing signatures. |
| `member_accessions` | Signature accessions from member databases. |
| `member_names` | Signature names from member databases. |
| `member_protein_counts` | Protein counts for member signatures. |
| `member_count` | Number of member signatures. |
| `external_databases` | External resource database names. |
| `external_accessions` | External resource accessions. |
| `external_xrefs` | Combined external cross-references as `DB:ACCESSION`. |
| `external_xref_count` | Number of external cross-references. |
| `pdb_ids` | PDB structure identifiers. |
| `structure_count` | Number of PDB structure links. |
| `publication_ids` | InterPro publication identifiers. |
| `pubmed_ids` | PubMed identifiers, when available. |
| `publication_titles` | Publication titles. |
| `publication_years` | Publication years, with `0` used when missing. |
| `publication_count` | Number of publications attached to the entry. |
| `parent_ids` | Parent InterPro entries from the XML hierarchy. |
| `child_ids` | Child InterPro entries from the XML hierarchy. |
| `parent_count` | Number of parents. |
| `child_count` | Number of children. |
| `tree_depth` | Minimum hierarchy depth from `ParentChildTreeFile.txt`, when present. |
| `taxonomy_names` | Taxa listed in the taxonomy distribution. |
| `taxonomy_protein_counts` | Protein counts for `taxonomy_names`. |
| `taxonomy_count` | Number of taxonomy distribution rows. |
| `key_species_names` | Key species names listed for the entry. |
| `key_species_protein_counts` | Protein counts for `key_species_names`. |
| `key_species_count` | Number of key species rows. |
| `in_entry_list` | Whether the entry appears in `entry.list`. |
| `entry_list_type` | Entry type from `entry.list`. |
| `entry_list_name` | Entry name from `entry.list`. |
| `names_dat_name` | Entry name from `names.dat`. |
| `short_names_dat_name` | Short name from `short_names.dat`. |
| `split_bucket` | Deterministic split bucket from `sha256(interpro_id) % 10`. |

## Preparation

The normalization script used to create the Parquet files is included at `scripts/prepare_interpro_dataset.py`.