Update dataset card for public release

README.md

@@ -4,51 +4,123 @@ license: cc-by-sa-4.0
 tags:
   - biology
   - protein
+  - protein-sequences
   - uniclust
   - uniclust30
   - uniref30
   - msa
+  - multiple-sequence-alignment
+  - proteomics
+  - bioinformatics
 pretty_name: UniClust30 (UniRef30)
 size_categories:
   - 10M<n<100M
+task_categories:
+  - feature-extraction
+configs:
+  - config_name: default
+    data_files:
+      - split: train
+        path: "train-*.parquet"
+dataset_info:
+  features:
+    - name: cluster_id
+      dtype: string
+    - name: representative_id
+      dtype: string
+    - name: sequence
+      dtype: large_string
+    - name: sequence_length
+      dtype: int32
+    - name: sequence_xxh128
+      dtype: string
+    - name: num_aligned
+      dtype: int32
+    - name: a3m
+      dtype: large_string
+    - name: member_count
+      dtype: int32
+    - name: member_ids
+      sequence: string
+  splits:
+    - name: train
+      num_examples: 36293491
 ---
 
 # UniClust30 (UniRef30)
 
-Complete UniClust30 / UniRef30 dataset (release 2023_02) from the Söding Lab,
-converted from HH-suite A3M format to sharded Parquet.
+Complete [UniClust30 / UniRef30](https://uniclust.mmseqs.com/) dataset (release 2023_02) from the Söding Lab, converted from HH-suite A3M format to sharded Parquet. UniClust30 clusters UniProt sequences at 30% sequence identity and provides pre-computed multiple sequence alignments (MSAs), making it a key input for protein structure prediction (e.g. ColabFold, AlphaFold) and co-evolutionary analyses.
 
-## Contents
+**Part of the [ConvergeBio Protein Database Collection](https://huggingface.co/collections/ConvergeBio/protein-database)** — see also [UniRef100](https://huggingface.co/datasets/ConvergeBio/uniref100), [UniRef90](https://huggingface.co/datasets/ConvergeBio/uniref90), and [UniRef50](https://huggingface.co/datasets/ConvergeBio/uniref50).
 
-Each row represents a UniRef30 cluster with:
-- **cluster_id**: Cluster identifier (e.g. UniRef30 accession)
-- **representative_id**: UniProt accession of the representative sequence
-- **sequence**: Representative protein sequence
-- **sequence_length**: Length of the representative sequence
-- **sequence_xxh128**: xxHash-128 of the representative sequence
-- **num_aligned**: Number of sequences in the A3M multiple sequence alignment
-- **a3m**: Full A3M-formatted multiple sequence alignment for the cluster
-- **member_count**: Number of cluster members (from mapping file)
-- **member_ids**: List of all member UniProt accessions
+## Dataset Summary
 
-## Source
+| | |
+|---|---|
+| **Clusters** | 36,293,491 |
+| **Shards** | 629 |
+| **Source release** | 2023_02 |
+| **Source format** | HH-suite ffindex/ffdata A3M database + cluster mapping TSV |
 
-- **Origin:** [Uniclust / UniRef30](https://uniclust.mmseqs.com/)
-- **Release:** 2023_02
-- **Format:** Converted from HH-suite ffindex/ffdata A3M database
-- **Compression:** zstd
-- **Reference:** Mirdita et al., "Uniclust databases of clustered and deeply annotated
-  protein sequences and alignments", *Nucleic Acids Res.* 2017
+## Schema
+
+Each row represents one UniRef30 cluster with its representative sequence, full A3M alignment, and cluster membership.
+
+| Column | Type | Description |
+|--------|------|-------------|
+| `cluster_id` | `string` | Cluster identifier (UniRef30 accession) |
+| `representative_id` | `string` | UniProt accession of the representative sequence |
+| `sequence` | `large_string` | Representative protein sequence (uppercase amino acid alphabet) |
+| `sequence_length` | `int32` | Length of the representative sequence in residues |
+| `sequence_xxh128` | `string` | xxHash-128 of the sequence (hex, computed at build time) |
+| `num_aligned` | `int32` | Number of sequences in the A3M multiple sequence alignment |
+| `a3m` | `large_string` | Full A3M-formatted MSA for the cluster |
+| `member_count` | `int32` | Number of cluster members (from mapping file) |
+| `member_ids` | `list<string>` | All member UniProt accessions |
 
 ## Usage
 
 ```python
 from datasets import load_dataset
 
+# Stream without downloading everything
+ds = load_dataset("ConvergeBio/uniclust30", streaming=True)
+for row in ds["train"]:
+    print(row["cluster_id"], row["num_aligned"], row["sequence_length"])
+    break
+
+# Or load fully
 ds = load_dataset("ConvergeBio/uniclust30")
 ```
 
+### Working with A3M alignments
+
+The `a3m` column contains the full multiple sequence alignment in [A3M format](https://github.com/soedinglab/hh-suite/wiki#a3m-format) (a compressed variant of FASTA). You can parse it directly or use HH-suite tools:
+
+```python
+row = next(iter(ds["train"]))
+lines = row["a3m"].split("\n")
+headers = [l for l in lines if l.startswith(">")]
+print(f"Cluster {row['cluster_id']}: {len(headers)} sequences in MSA")
+```
+
+## Data Processing
+
+- **Source:** `UniRef30_2023_02_hhsuite.tar.gz` and `uniref_mapping.tsv.gz` from [uniclust.mmseqs.com](https://uniclust.mmseqs.com/)
+- **Parsing:** Read ffindex/ffdata A3M database; representative sequence and alignment count extracted from each A3M entry; cluster members from mapping TSV
+- **Integrity:** xxHash-128 computed per sequence; A3M-extracted sequence verified against stored sequence for every row
+- **Validation:** Passed all tiers — schema conformance, zero null/empty sequences, xxHash roundtrip, A3M format validation, member ID consistency against mapping TSV, field-by-field comparison against source ffindex/ffdata, and exact row count match (36,293,491)
+- **Format:** Sharded Parquet with zstd compression
+
+## Source & Citation
+
+UniClust30 is produced by the [Söding Lab](https://www.mpinat.mpg.de/soeding):
+
+> Mirdita M, von den Driesch L, Galiez C, Martin MJ, Söding J, Steinegger M.
+> "Uniclust databases of clustered and deeply annotated protein sequences and alignments."
+> *Nucleic Acids Research* 45(D1):D206–D213 (2017).
+> [doi:10.1093/nar/gkw1081](https://doi.org/10.1093/nar/gkw1081)
+
 ## License
 
-UniClust30 data is available under
-[CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/).
+UniClust30 data is available under [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/).