chewcall: fast, chewBBACA-compatible cgMLST/wgMLST allele calling with exact Smith-Waterman and a schema-side filter audit

Résumé fourni par la source

Reproducibility data accompanying the paper “chewcall: fast, chewBBACA-compatible cgMLST/wgMLST allele calling with exact Smith-Waterman and a schema-side filter audit”. This deposit contains the allelic profile matrices produced by chewcall and chewBBACA on the benchmark datasets, together with the analysis scripts needed to regenerate every table and figure of the paper. It does not include the schemas or genome assemblies (these are large and already publicly available); it contains the outputs of allele calling, which are what the analysis scripts consume. Companion code: https://github.com/genpat-it/chewcall (tag v0.3.0). Two levels of reproduction Level 1 — regenerate all tables/figures (this deposit + Python). The scripts in reproduce/ read the profile matrices in results/ and emit the LaTeX tables and figures. Requires only python3 with numpy, pandas and scipy — no chewcall, no chewBBACA, and no schemas or genomes are needed. Level 2 — re-run the tools from raw genomes (optional). To regenerate the matrices themselves, additionally obtain (all public): chewcall (build from the repository or docker pull ghcr.io/genpat-it/chewcall:0.3.0); chewBBACA v3.5.4 / v3.3.10 (via Bioconda); the genome assemblies (BeONE Zenodo deposits cited in the paper); and the schemas (Chewie-NS). Contents results/cc/ — chewcall allelic profile matrices (shared-CDS and end-to-end). results/c354/ and results/c3310/ — chewBBACA v3.5.4 and v3.3.10 reference matrices. cgmlst_loci/ — cgMLST locus-name lists for S. enterica, E. coli, C. jejuni. fda/ — FDA Gen-FS Gopher outbreak profiles (chewcall and chewBBACA), plus fda/metadata.tsv listing the NCBI SRA accession, strain and outbreak/outgroup role of every sample. s2_genomes/ — one genome assembly per organism, for the Supplementary S2 pyrodigal-equivalence check. reproduce/ — Makefile, analysis scripts and README (paths are archive-relative). Coverage. Every table and figure in the paper is reproducible. From this archive plus Python alone you can regenerate Tables 1–8, Figure 3, and Supplementary S1/S2/S3/S4/S5. The minimizer-parameter ablation (S4.1) and the speedup-decomposition ablation (S7) instead require re-running chewcall (the latter also BLASTp) on the BeONE data, whose genomes (BeONE Zenodo deposits) and schemas (Chewie-NS) are publicly available as described above. Each results/ / / directory holds results_alleles_hashed.tsv (CRC32 hashes of the called DNA, for sequence-level concordance and clustering) and results_alleles.tsv (allele and class labels, for the actionable-difference and classification tables); timing_*.txt files carry the wall-clock measurements for the runtime tables. See reproduce/TRACEABILITY.md for the full artefact → script → value mapping. Datasets. Eight BeONE benchmark datasets (4 organisms × consortium and public collections: Listeria monocytogenes, Salmonella enterica, Escherichia coli, Campylobacter jejuni) plus four FDA outbreak datasets. License: data CC-BY-4.0; the accompanying source code is released under GPL-3.0 in the linked repository.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.