Reproducibility package for: Auditing bipartite motif interpretations (analysis scripts, cached ensembles, figure code)

Abstract

This package reproduces the analyses, cached null ensembles and figures of "Auditing bipartite motif interpretations: a worked example with conservation checks and open-path decomposition" (Tengfei Shao, submitted to PeerJ Computer Science, 2026). It contains the typed bipartite motif counting operator and three independent cross-checks of it; the degree-preserving null pipeline that draws the record and robustness ensembles used throughout the note; the chain-length convergence run and its second-seed replicate; the leave-one-tourist-out, drop-two-most-active and edge-rating-threshold robustness re-runs; the single number source (numbers.csv) with the reconciliation scripts; and the eight figure scripts with their rendered output. MANIFEST.json carries the SHA-256 and byte count of every file. The tourism rating matrix that every script operates on (17 tourists x 80 sightseeing sites) is openly available (CC BY 4.0) at 10.5281/zenodo.22299150 and is not bundled here; data/README_DATA.md explains how to obtain it and verify it by MD5 hash. Every script accepts a TOURISM_CSV environment variable so the file can be placed anywhere. The null draws depend on attrimotif 1.1.1 (10.5281/zenodo.22259239); every script accepts an ATTRIMOTIF_SRC environment variable pointing at a source checkout or falls back to an installed package. This package covers only the note's own analyses. It is not a re-release of the predecessor framework package (10.5281/zenodo.22287908, version 5, corrected proposal-counted sampler), which the note applies to a single worked example. Run order, expected key outputs, seeds and environment are given in README.md. Licence: MIT for code; CC BY 4.0 for result tables and figures. Version 1.6 (2026-09-05). Table numbers in README.md remapped to the sequential Tables 1 to 10 used by the manuscript. The manuscript renumbered its tables on 2026-09-05, from the earlier 1a/1b/1c/2/3/4a-4d/5 scheme, so that the submission portal's own table slots match the captions; the mapping is 1a to 1, 1b to 2, 1c to 3, 2 to 4, 3 to 5, 5 to 6, 4a to 7, 4b to 8, 4c to 9, 4d to 10. No code, data, figure or result changed between 1.5 and 1.6. Version history: 1.5 tourism-matrix references updated to the current record version (10.5281/zenodo.22299150) and the Figure 5 subtitle recalibrated ("extreme relative to the sampled null" instead of "real under the null"); 1.4 aligned version labels and the cross-check header and removed the regression-based conditional z from Figure 5; 1.3 README mapping statement; 1.2 open-data notes, TOURISM_CSV, mappings, pinned requirements, licence scope. No analysis result changed in any version. Funding: This work was supported by the Waseda University Grant for Special Research Projects. This work was supported by MEXT Supporting Pioneering Research through AI for 1,000 Discovery challenges Program (SPReAD) Japan Grant Number JPMXP1726306059.