# piRNAome 2026.1 — data release

**piRNAome: a harmonized, isoform-aware and locus-resolved piRNA database across five species**
Version **2026.1** · Zitong Gao (University of Hawai‘i at Mānoa)
Web portals: https://pirnaome.org (human) · https://species.pirnaome.org (other species)
License: **CC-BY-4.0**

PIWI-interacting RNAs (piRNAs) are 24–31-nt small RNAs involved in transposable-element silencing and gene
regulation. piRNAome is an evidence-preserving database that harmonizes seven major sources and organizes
publication aliases, mature sequences, isoform families, verified genomic placements and associated
annotations within a unified framework, for *Homo sapiens*, *Caenorhabditis elegans*,
*Drosophila melanogaster*, *Mus musculus* and *Rattus norvegicus*.

## Packaging
One archive per species — `<sp>_piRNAome_2026.1.tar.gz` — each containing the folder `<sp>/` (7 data files +
`CHECKSUMS.md5`) plus this README. Species (folder / assembly): **hsa** = *H. sapiens* (GRCh38/hg38) ·
**cel** = *C. elegans* (ce11/WBcel235) · **dme** = *D. melanogaster* (dm6/BDGP6) ·
**mmu** = *M. musculus* (**mm10**/GRCm38) · **rno** = *R. norvegicus* (rn6/Rnor_6.0).

Each `<sp>/` folder contains (`.gz` = gzip-compressed):

| file | description |
|---|---|
| `<sp>_piRNA.fasta.gz` | all mature piRNA sequences; header `>ACCESSION display_id=…;length=…` |
| `<sp>_piRNA.gtf.gz` | one feature per piRNA at its **primary genomic locus**; coordinate-less piRNAs excluded |
| `<sp>_piRNA_clusters.gtf` | piRNA clusters — **known + de novo** |
| `<sp>_id_aliases.tsv.gz` | source-database identifier aliases per piRNA |
| `<sp>_families_annotation.tsv[.gz]` | isoform families + annotation |
| `<sp>_family_overlaps.tsv[.gz]` | pairs of **overlapping / adjacent piRNA families** on the same strand (renamed from “junctions” in 2026.1) |
| `<sp>_snv_results.tsv[.gz]` | locus-level sequence-variation records (human); coordinate-verification mismatches (other species) |

## Conventions
- **Public ID = accession**: piRNA `piRO-<sp>-P########`, family `…-F########`, locus `…-L#########`, cluster `…-C########`.
- **Coordinates**: GTF files are **1-based, inclusive** (DB is 0-based half-open → `start+1`).
- **Chromosome names**: **canonical UCSC** per assembly (e.g. mm10 `chr1…chrM`, `chr*_random`, `chrUn_*`).
  Other-species coordinates were normalized to canonical UCSC names (from a mix of UCSC/Ensembl/GenBank) in
  Aug 2026; human GRCh38 was already UCSC-canonical.
- **Strand**: `+`/`-`; `.` when unknown/bidirectional.

## File details
**piRNA.fasta** — every piRNA in the catalogue (not only genome-placed ones).

**piRNA.gtf** — feature `piRNA`, one line per piRNA at its primary genomic locus. piRNAs with **no genomic
placement are omitted** (human: `loci__n.is_primary_locus`; other species: excludes `coord_status` ∈
{NO_GENOMIC_MATCH, NO_FULL_LENGTH_CANDIDATE, CANDIDATES_EXCEED_2MM} and coordinate-less realignments).
Attributes: `piRNA_id`, `locus_id`, `locus_status`/`coord_status`, `genomic_context`, and (when assigned)
`family`, `known_cluster`/`cluster`, `denovo_cluster`. Multi-mapping piRNAs are represented by their primary
locus; all placements remain in the database.

**piRNA_clusters.gtf** — feature `piRNA_cluster`; attributes `cluster_id` (accession), `cluster_type`
(`known`|`de_novo`), `n_piRNA`, and for de novo `tier` + `poisson_pvalue`.

**id_aliases.tsv** — `piRNA_accession`, `source_db`, `original_id`, … Source DBs: piRBase, proTRAC, piRNAdb,
piRNABank, piRNAQuest, NCBI, piOxiDB.

**families_annotation.tsv** — family accession, member/placement counts, anchor/root context and dominant
TE / gene annotation.

**family_overlaps.tsv** — pairs of piRNA loci/families that **overlap or are adjacent on the same strand**
(relabeled from “junction” in this release). Human: `junction_key`, `chromosome`, `strand`, `gap_nt`,
five′/three′ locus ids, both partners’ piRNA/locus/family accession + legacy id + sequence + `locus_status`.
Other species: `junction_id`, `species`, `chromosome`, `strand`, `gap_nt`, `overlap_class`, `status`,
`evidence_reads`, and for each partner (a/b) the piRNA accession, locus id, coordinate, anchor sequence.
*(The internal id column is still named `junction_key`/`junction_id`.)*

**snv_results.tsv** — Human: locus-level SNVs with dbSNP/ClinVar/COSMIC/REDIportal cross-references. Other
species: genome-vs-piRNA mismatches from coordinate verification (`genome_snv`); dbSNP/ClinVar/COSMIC/
REDIportal are human-only (cel file is header-only).

## Row counts

| species | piRNAs (FASTA) | placed loci (GTF) | clusters (known+de novo) | family overlaps | aliases | families | SNVs |
|---|--:|--:|--:|--:|--:|--:|--:|
| hsa | 10,033,305 | 9,942,083 | 14,487 (1,083 + 13,404) | 785,144 | 11,315,648 | 4,586,806 | 4,952,525 |
| cel | 30,025 | 30,025 | 160 (40 + 120) | 1,335 | 45,929 | 26,355 | 0 |
| dme | 87,186 | 86,385 | 1,185 (252 + 933) | 8,367 | 89,980 | 40,704 | 2,480 |
| mmu | 132,153 | 131,279 | 1,764 (1,092 + 672) | 10,381 | 139,826 | 59,771 | 3,290 |
| rno | 56,705 | 56,459 | 224 (137 + 87) | 7,375 | 56,987 | 34,953 | 490 |

## Provenance
Generated 2026-08-05 from `pirnaome_human_2026.1.sqlite` (+ `cluster_stats_2026.1.sqlite` for known-cluster
coordinates) and `piRNA_otherspecies_2026.1.db` (opened read-only). In Aug 2026 the other-species chromosome
names were normalized to canonical UCSC and the piRNA GTFs re-emitted (coordinate-less features dropped:
dme 189, mmu 158, rno 10); the “junction” files were renamed to “family_overlaps”. `export_human.py` /
`export_otherspecies.py` (original generation scripts) are included for provenance; the normalization above
was applied as a separate post-processing step. Integrity: `CHECKSUMS.md5` (MD5) inside each species folder.

## Citation
Gao, Z. *piRNAome: a harmonized, isoform-aware and locus-resolved piRNA database across five species* (Version 2026.1) [Data set]. Zenodo. DOI: *reserved on publication*.