vivarium-preplisted

# vivarium-prep — assembly QC and annotation Take raw reads or draft genomes to the state the rest of vivarium needs: a quality-assessed assembly and gene/function annotation. Fast QC runs now in `bio_tools`; slow steps (Flye assembly, eggNOG, dbCAN) are handed back as exact commands so they don't block the session. ## Step 1 — where are we in the pipeline? ``` reads --(assembly: Flye, heavy)--> contigs --(QC: stats + CheckM2)--> genome --(annotation: Prokka -> eggNOG/dbCAN)--> features ``` Most users arrive with a genome already and want **QC + annotation**. Start there unless they have raw reads. ## Step 2 — assembly QC (bundled, fast) ```bash bash <skill-dir>/scripts/prep.sh stats --genome <genome.fna> --out genome_stats.tsv # or a whole folder: --indir <genomes_dir> (one row per genome -> feeds vivarium-report bars) ``` Reports per genome: contigs, total length, N50, GC%, longest contig, gaps. A closed bacterial genome is 1–few contigs with N50 ≈ genome size; hundreds of contigs / low N50 means a fragmented draft — say which, because it changes how much to trust downstream synteny and unique-gene calls. **Completeness/contamination (CheckM2)** is the other half of QC but is *not* installed in this env, so scaffold it: ```bash checkm2 predict --input <genomes_dir> --output-directory checkm2_out -x fna --threads <N> ``` Read `quality_report.tsv` → Completeness ≥ ~95% and Contamination ≤ ~5% is a sound MAG/isolate; flag anything below. ## Step 3 — assembly (Flye,