Plant- and invertebrate-associated Lactobacillaceae from public 16S rRNA gene amplicon data: classified operational taxonomic units, host resolution, analysis frames and fitted models

Derived outputs of a survey of Lactobacillaceae in plant and invertebrate hosts, mined from 948,386 public 16S rRNA gene amplicon runs in the Sequence Read Archive. Every quantity the article's figures and tables report can be checked from these files without repeating the retrieval, which is approximately 28 TB. Eleven groups: the per-run recruitment outcome for every run retrieved, the control-excluded occurrence stream and the three per-run covariate tables, the three classifier stages, the rank calls, the per-OTU amplicon region, identity to the nearest described type strain, the per-run host and habitat resolution, the detections table and both effort frames, the genome tree of record and its core-gene alignment, the fitted Firth models, and the deposited tables. Every group carries the provenance file its builder wrote, recording the SHA-256 of each input read. The deposit is sized so that the analysis code archived alongside it runs end to end from these files: its first step reads the occurrence stream, and every input from there to the reported tables is present. Raw reads are not included. They are public in the SRA, and the accession lists that define what was retrieved ship with the analysis code. The data travel as one gzipped tar of 187 members, with README.md, LICENSE and MANIFEST.sha256 also placed beside it so the contents, the licence and the checksums can be read without downloading 1.1 GB. MANIFEST.sha256 covers all 184 data files and is inside the tarball as well, so sha256sum -c MANIFEST.sha256 verifies every one of them from the extracted root with no further download; SHA256SUMS.txt carries the checksum of the tarball itself. Files above 20 MB are gzipped with no timestamp, and the tarball pins member order, ownership and every mtime, so the deposit is byte-reproducible.