TAMP is a modular, automated pipeline for benchmarking, selecting, and merging genome assemblies. It is optimized for fungal haploid genomes and PacBio HiFi reads, but is generally applicable to small eukaryotic genomes.

• Runs multiple assemblers (1–6): HiCanu, NextDenovo, Peregrine, IPA, Flye, RAFT-hifiasm.
• Normalizes and copies assemblies (7): consistent contig headers and length-sorted FASTA per assembler.
• Telomere discovery & per-assembler telomeric subsets (8–9): finds telomeric contigs (via seqtk telo) and prepares per-assembler telomeric FASTAs.
• Quality assessment (10, 13): QUAST for structure metrics; BUSCO for completeness.
• Unified assembly metrics (12): before merging, TAMP rebuilds assemblies/assembly_info.csv by combining BUSCO/QUAST/TELO CSVs so you can compare candidates in one table.
• Final merge with T2T protection (12):
  • You choose an assembly interactively or with --choose.
  • T2T contigs from t2t_clean.fasta are always preserved.
  • Redundancy reduction (redundans) runs only on non‑T2T contigs, with an optional minimap2 filter to drop non‑T2T contigs redundant to T2T.
  • Contigs are recombined and length‑sorted for the final FASTA.
• Telomere QC & final summary (14–17): telomere counts, telomeric FASTA, and consolidated reports.
• Reproducibility: each run writes tool versions to version.txt and step-by-step logs to logs/.

conda env create -f dependency/pacbiohifi.yml
conda env create -f dependency/busco.yml
conda env create -f dependency/quast.yml

TAMP activates the appropriate environment automatically during steps.

Full run (HiFi reads):

bash TAMP-0.2.7.sh --fastq reads.fastq.gz -g 90m -t 20 -m AACCCT --busco ascomycota_odb10

Resume from Step 7:

bash TAMP-0.2.7.sh --fasta genome.fa -g 90m -t 20 -m AACCCT -s 7-17

Run merge only (Step 12+) with a prebuilt T2T:

bash TAMP-0.2.7.sh -g 90m -t 32 --fasta myassembly.fa -s 12-17
# Non-interactive choose:
bash TAMP-0.2.7.sh -g 90m -t 32 --fasta myassembly.fa -s 12 --choose flye

1. Rebuild summary table first
   At the start of Step 12, TAMP rebuilds assemblies/assembly_info.csv by merging any of these that exist:

   • assemblies/assembly.busco.csv (or common alternates)
   • assemblies/assembly.quast.csv
   • assemblies/assembly.telo.csv
     The builder is Python‑based in v0.2.7+, so it’s robust to CRLF line endings and awk variants.

2. Choose an assembly

   • By default (v0.2.7), Step 12 auto-selects the assembler with the highest N50 in assemblies/assembly_info.csv.
   • You can override this with --choose <assembler>.
   • If auto-selection cannot decide (e.g., missing N50 row), an interactive prompt lists assemblies present in assemblies/*.result.fasta.

3. T2T protection and merge

   • All contigs from t2t_clean.fasta are protected and kept as‑is.
   • The chosen assembly’s non‑T2T contigs are reduced with redundans (T2T is never passed into redundans).
   • (Recommended) minimap2 -x asm20 can drop non‑T2T contigs that are redundant vs. T2T (e.g., identity ≥ 0.95; covered fraction ≥ 0.95).
   • Protected T2T + reduced “others” → recombined and length‑sorted → assemblies/final.merged.fasta.

• Double‑end contig: a contig with telomeric signal at both ends (has at least one hit where start == 0 and at least one hit where end == length).
• Single‑end contig: telomeric signal at exactly one end.

These definitions are enforced in v0.2.2+ to avoid under‑counting double‑ended contigs.

• assemblies/*.result.fasta — normalized per‑assembler FASTAs
• assemblies/final.merged.fasta — final assembly
• assemblies/assembly_info.csv — unified matrix of BUSCO/QUAST/TELO metrics
• assemblies/merged.telo.csv — final telomere metrics table
• assemblies/merged.busco.csv — final BUSCO metrics table
• assemblies/merged.quast.csv — final QUAST metrics table
• merged_result.csv — consolidated comparison of assemblies (from Step 16)
• quast_final/ — QUAST outputs
• logs/step_<N>.log — logs per step
• version.txt — tool versions captured during the run

• No assembler found / empty files: check paths and that each assembler completed.
• Step 12 table looks empty: confirm at least one of BUSCO / QUAST / TELO CSVs exists in assemblies/.
• Seqtk not found: ensure the funannotate (or appropriate) env is active; TAMP will try to auto‑activate.
• Line‑ending issues: v0.2.7+ Python builder strips CRs; previous awk errors (e.g., gsub(/ at Step 12) are resolved.
• T2T contig names changed by a merger: use the alignment‑based T2T split path to protect T2T contigs by alignment rather than name.

If you use TAMP v0.2.7, please cite:

Sun, Y. (2025). TAMP: Telomere Assembly Merge Pipeline v0.2.7.
Grainger Bioinformatics Center, Field Museum of Natural History.

• v0.2.7 — Final merge & summaries: Auto-selects the assembler with the highest N50 in Step 12 (unless --choose is given); runs redundans with minimap2-based reduction on non‑T2T contigs; writes telomere/BUSCO/QUAST summary tables as assemblies/merged.*.csv and a consolidated merged_result.csv.
• v0.2.6 — Step 12 builder hardened: Replaced awk table builder with Python‑based build_assembly_info_v2 (CRLF‑safe, header‑normalized) and invoked it at the start of Step 12.
• v0.2.5 — Step 12 robustness: Made CR removal explicit to avoid awk /.../ parse errors on some platforms.
• v0.2.4 — Step 12 flow: Always rebuild assemblies/assembly_info.csv from BUSCO/QUAST/TELO prior to prompting/--choose; print the matrix.
• v0.2.3 — Step 7 bugfix: Removed stray Bash call embedded in the Python heredoc of rename_and_sort_fasta (no more SyntaxError in /tmp/rename_fa*.py).
• v0.2.2 — Telomere counts: Fixed double‑end logic in Steps 9 & 14 to require signal at both ends; single‑end unchanged.
• v0.2.1 — T2T protection: Step 12 ensures T2T contigs are preserved; run redundans only on non‑T2T contigs; keep --choose and interactive prompt.