TRANSDECODER(1)

👑 NAME

TransDecoder — the full pipeline wrapper, bestie. it does the whole thing for you, start to finish 💖

---

🌸 DESCRIPTION

okay so picture this: you've got your sequences, you've got dreams, and you do not want to run five separate tools by hand like it's 2009. TransDecoder is literally the wrapper that runs the whole pipeline in one go, in this exact order:

LongOrfs -> [optional homology search] -> Predict -> [optional genome propagation]

periodt. that's the whole vibe. ✨

🎀 Input (choose one)

you have to pick one way to feed it, and i mean it, choose one:

• -t|--transcripts <string> — the classic: a Transcripts FASTA file. just give it your transcripts and go 🫶
• --genome <string> / --gtf <string> — the genome-flavored route, and these two are a pair, ok? you use them together so it can extract cDNA sequences via gtf_genome_to_cdna_fasta.pl, and then propagate the final ORFs back to genome coordinates. --genome is your Genome FASTA file, --gtf is your Annotation GTF file. iconic teamwork. 💫

🧬 LongOrfs options

the first stage, where it hunts down the ORFs:

• -m <int> — Minimum protein length (default: 100). tiny proteins need not apply 💀
• -S — Strand-specific (top strand only). only looks at the top strand, no cap.
• -G|--genetic_code <string> — Genetic code (default: universal). change it if your organism is built different.
• --gene_trans_map <string> — Gene-to-transcript map (tab-delimited). tabs only, bestie, not spaces 🙈
• --complete_orfs_only — Only report complete ORFs. no half-finished situationships.

🔥 Homology search options

this part is optional — it only kicks in when you ask for it:

• --blast_search_pep <string> — Protein FASTA to search against; this is the thing that triggers the homology search, and the DB gets built automatically for you. you don't lift a finger 💅
• --blast_tool <string> — 'diamond' or 'blastp' (default: diamond). diamond is the default because it's fast, we love her.
• --blast_evalue <float> — E-value cutoff (default: 1e-5). yes, 1e-5, that's the number.
• --blast_threads <int> — Threads for homology search (default: 1). crank it up if your machine can handle the drama.

🌈 Predict options

the glow-up stage:

• -T <int> — Top ORFs for Markov model training (default: 500).
• --retain_long_orfs_mode <string> — 'dynamic' or 'strict' (default: dynamic).
• --retain_long_orfs_length <int> — Min length to auto-retain under strict mode. only does its thing when you're in strict, iykyk.
• --pfam-search-db <string> — Pfam HMM database to search with hmmsearch; hmmpress is run automatically if needed. yes, automatically. obsessed. 😭
• --single_best_only — Retain only single best ORF per transcript. no backups, no roster, just the one.
• --no_refine_starts — Skip start codon refinement. just skip it, live your truth.

💜 Other

the little extras:

• -O|--output_dir <string> — Output directory (default: current directory). it'll just dump everything right where you are if you don't say otherwise.
• -v|--verbose — Verbose output. for when you need the tool to narrate.
• --version — Show version and exit. the quickest exit ramp, no notes.

---

⚠️ heads up bestie: when it says choose one input, it means it. the --genome + --gtf combo travels together, and --blast_search_pep is the only thing that turns the homology search on. don't come for me if you skip that. 🥲

---

that's the whole page, we're done here 🎀 this was TransDecoder 6.0.0, dated May 2026 — go find some ORFs, you've got this ⭐