Public stub for Vitis TE libraries and the soft-mask hand-off into gene annotation.
Operational notebook: private lab TE tree (not in git).
Pipeline idea: EDTA → TEtrimmer → TEsorter → CDS-gated trusted → soft-mask / panEDTA.
R1 numbers (e.g. trusted ~240) are archive/teaching examples; panel gold may move to a newer trusted emit.
Gene soft-mask (A0) only accepts the current trusted file + version/sha256 — not raw EDTA / whole working.
Structure playbook plug-in:
gene-structure-annotation docs/TE_LIBRARY.md.
- Not gene-model annotation — gene-structure-annotation
- Not functional annotation — gene-function-annotation
- Not PAV graphs — vitis-pangenome
- Not a dump of another group’s raw TE calls
- EDTA per assembly/haplotype (
--species others --sensitive 1 --anno 1; document--u) - TEtrimmer per hap → merge Perfect/Good → CD-HIT ~95% working lib
(optional 80-80 family catalog; do not overwrite working) - TEsorter (e.g. rexdb-plant) — domains/labels only; never replace working FASTA with
all.cls.lib - Curation gate (CDS BLAST + TEsorter class emit rules) → trusted FASTA (record sha256)
- Soft-mask / EDTA
--curatedlibwith trusted only - panEDTA combine (official; not
catof TElibs) → panel reannotate - LTR age / LAI / PAV after curated pan TE GFFs
| Product | Use as --curatedlib / gene soft-mask? |
|---|---|
| Raw EDTA TElib | No |
| Working lib (post-TEtrimmer CD-HIT) | No (whole file) |
| Trusted gated lib | Yes |
cat + CD-HIT of many libs |
No (dedup ≠ curation) |
EDTA.pl --genome assembly.fa --species others --sensitive 1 --anno 1 --threads 32- Do not treat EDTA raw output as a gold-standard TE library.
- Do not treat
cat+CD-HIT as curatedlib. - Classifier-only tools name consensi; they do not replace trim or become the library.
- Do not hard-mask for BRAKER/GALBA; soft-mask with a trusted, host-gene-purged lib.
- Non-TE repeats (TRF / telomere / rDNA) are not curatedlib material.
Author: Xuzhen Li · ORCID