Files uploaded to this Dryad dataset

The Dryad submission contains two uploaded ZIP files, with the following exact filenames:

The main archive includes README.md, FILE_INVENTORY.csv, VARIABLE_DICTIONARY.csv and CSV_MISSINGNESS_AUDIT.csv. Together they provide the folder guide, exact internal file listing, tabular variable definitions and missing-value explanations. The main ZIP and its internal root folder share the name Dryad_Submission_311026_14 to simplify navigation; BV-BRC_Annotations.zip is a separate uploaded file.

Submitted archive name and layout

The authoritative data-and-code archive for this Dryad deposit is Dryad_Submission_311026_14.zip, containing the top-level folder Dryad_Submission_311026_14/. All relative file paths in FILE_INVENTORY.csv, VARIABLE_DICTIONARY.csv, and CSV_MISSINGNESS_AUDIT.csv begin inside this folder unless a nested ZIP is identified using !. The file BV-BRC_Annotations.zip is not inside this archive; it is uploaded separately to the same Dryad dataset and its contents are listed in BV-BRC_annotation_contents.csv. Earlier package/version names are superseded by this archive and should not be uploaded as additional copies.

The folder DRYAD_R_script_standardised_bundle/ contains a deliberate duplicate of the main root-level R script. Technical README files inside nested software/analysis bundles remain as local instructions; README.md at the archive root is the authoritative dataset-level documentation.

Dataset scope and provenance

This dataset accompanies the accepted manuscript on Pleistodontes fig-pollinating wasps and Wolbachia cytoplasmic incompatibility (CI). Its principal identifiers are the JRAS specimen codes. The data-and-code package contains sequence alignments, Wolbachia assemblies, phylogenetic trees, capture-bait sequences, CI comparison tables, structural models, electrostatic/contact analyses, plotting inputs and R/Python scripts. The large BV-BRC genome and phage annotation archive is supplied as a separate file in the same Dryad dataset, named BV-BRC_Annotations.zip; it is documented by the accompanying BV-BRC_annotation_contents.csv manifest. The archive is not duplicated inside this data-and-code ZIP.

Deposit structure: This ZIP contains the data and code documented below. The full BV-BRC_Annotations.zip archive is a separate upload in the same Dryad dataset; its filenames are listed in BV-BRC_annotation_contents.csv.

Interpretation and scope of inferred reproductive compatibility

The cif-based compatibility groups, pairwise cytoplasmic incompatibility (CI) comparisons, and structural/electrostatic scores in this deposit are computational inferences, not CI phenotypes verified through controlled crosses. They draw on recovered gene sequences, protein structure and charge-complementarity analyses; their classification should not be treated as direct experimental evidence of failed or successful reproduction. The original37 (37 hotspot-pair) analysis underlies the principal electrostatic comparison presented for the manuscript. The deposited 45-scaffold, candidate-partner, partial-protein, validation and other structural analyses are supporting or exploratory analyses with distinct inclusion criteria, not alternative interchangeable primary results. See the corresponding scripts and the manuscript for their scopes and assumptions.

A negative or unavailable Wolbachia, MLST or cif assignment can reflect assay detection limits, incomplete sequence recovery, insufficient reads or the absence of an applicable comparison. Interpret W-, noWol, 0, NA, - and an empty cell according to their specific file and variable definitions; they are not mutually interchangeable. In particular, an uncovered residue in a partial protein comparison is not equivalent to an unchanged residue. These distinctions are reflected in VARIABLE_DICTIONARY.csv and PROTEIN_MISSING_VALUE_CODING_REPORT.csv.

Dataset navigation and source-output provenance

Start with Summary Data Table.csv for specimen-level context; use the principal MLST/cif tables and the Protein Poses and Functional Residues.zip 7_the_summary_tables/csv/ outputs for curated summaries. Files within Primary Analysis/, supporting-analysis and validation directories retain the computational history and alternative assessments and should be interpreted according to their original scripts. The root-level Analysis_Script_Pleistodontes_Wolbachia_DRYAD_standardised.R is intentionally duplicated within DRYAD_R_script_standardised_bundle/; these are not separate analyses. Nested software instructions are retained as technical documentation and do not supersede this dataset-level README.

PHASTEST, BV-BRC and wolPredictor software-generated output fields have been preserved with their native names and conventions for traceability. Some such fields are conditional, not reported, or inapplicable to particular records; blanks and NA values do not imply numerical zero or biological absence. For the protein analysis specifically, PROTEIN_MISSING_VALUE_CODING_REPORT.csv describes the grounds for explicit NA and - coding by output and column. The original PHASTEST files are unaltered and their blank annotation fields are described below. The 47 original BV-BRC .xls exports remain in the separately submitted annotation archive.

Manuscript figure exports and retained analytical graphics

Rendered PDF/PNG figure exports were removed from cifB Phylogeny/ and Phylogeny Figures SI/, because these visualisations appear in the accepted manuscript or supplementary information. The Phylogeny Figure/ folder had already had its R-generated visual exports removed; it now retains its tree and trait input data. The sequence alignments, Newick trees, phylogenetic input tables and scripts remain available, so plots can be regenerated from deposited data. A consolidated record of all removed rendered files is supplied as REMOVED_RENDERED_FIGURES.csv. These removed figures are not part of this deposited archive; please consult the journal article for the published visual versions.

To avoid republishing potentially journal-controlled manuscript figures under Dryad's CC0 waiver, 14 figure exports directly generated by Analysis_Script_Pleistodontes_Wolbachia_DRYAD_standardised.R were removed from this data-and-code archive. These comprised two cophylogeny renderings, two host-phylogeny/trait renderings and ten geographical map renderings in PDF/PNG/JPG/SVG formats. The underlying CSV/tree input files, R plotting code, analysis scripts and related non-rendered data were retained, so the visualisations can be regenerated where permitted. The removed exports are listed in REMOVED_R_FIGURES.csv.

The remaining rendered graphics in this data-and-code archive are analytical plots inside wolPredictor_240526.zip (including prediction and null-model outputs). The bait-screening report PDF is not included in this deposit. The sequencing-provider report is not included in this deposit. The separately uploaded BV-BRC_Annotations.zip is documented by its member inventory, rather than by inspection of each annotation output.

How to navigate the data

• FILE_INVENTORY.csv lists the files in this data-and-code package, including each member of the nested ZIPs, and identifies BV-BRC_Annotations.zip as a separate file in the same Dryad dataset. The annotation archive's 1,151 internal members are listed in BV-BRC_annotation_contents.csv.
• VARIABLE_DICTIONARY.csv lists every column of every CSV/TSV, including CSVs inside the nested ZIP files, with a working definition, observed sample values, whether units/keys are known and number of blank cells.
• CSV_MISSINGNESS_AUDIT.csv lists which tabular columns contain blank cells, for inspection by curators/users.
• CONVERSION_LOG.csv lists the four Excel-to-CSV conversions. Data were transferred as cell values; Excel-only styling, comments, filters, merges, freeze panes and visual effects are not carried into CSV.
• BV-BRC_annotation_contents.csv lists all 1,151 members of the separately uploaded annotation archive, including their internal paths, extensions and sizes.

Included file groups and interpretation

For folder-level examples and analysis provenance, see Detailed guide to folders, key files and analysis workflows below.

Detailed guide to folders, key files and analysis workflows

The descriptions below expand the navigation table with examples of actual deposited files. Some folders include both analytical inputs and outputs to support inspection and reuse; not every subworkflow is a single-command reproduction pipeline. The identifiers in filenames should be retained when matching records between datasets. For a complete file-by-file list, including nested ZIP members, consult FILE_INVENTORY.csv.

Specimen summary and genome assemblies

• Summary Data Table.csv provides specimen-level metadata and should be the first stop when matching JRAS identifiers across data groups.
• Assemblies.zip contains 48 specimen-coded Wolbachia genome assembly FASTAs. The source FASTA filenames sometimes end in NW, NP, WB, NA or P. The original project documentation used these suffixes for no Wolbachia, no phage, supergroup B Wolbachia, too few reads, and phage, respectively. These historical labels should be interpreted alongside specimen-level evidence, not as independent proof of biological absence.

Comparative infection patterns, phylogenies and geographic plots

• CI Comparison/ contains exaData_faoReview.csv, the main global-versus-within-host compatibility comparison summaries and the corresponding binomial and proportion-test results. Filenames distinguish the MLSTU and cifProteins classification schemes. These comparisons use inferred, not crossing-experiment-validated, CI categories.
• cifB Phylogeny/ provides cifB_Multispecies.fasta, the rooted and unrooted IQ-TREE tree files and JRAS_vs_reference_cifB_pairwise_identity.csv. The latter is a pairwise sequence-comparison output, not a direct estimate of reproductive compatibility.
• Cophylogenies/ contains the host, Wolbachia, cifA and phage tree files and label_colours.csv. Rendered cophylogeny figure exports have been removed; these remaining files are inputs for host–symbiont/tree comparisons and associated PACo analyses.
• Map Figure/ includes Map Figure.csv with coordinates, host identifiers and compatibility-group plotting columns, with published-style map exports removed from this deposit. Use the README's host and colour keys when interpreting these plotted groups.
• Haine and Cook (2005) Data/ contains a comparison alignment and phylogenetic tree independently reconstructed from published sequence data. These files do not reproduce figures from the original publication.
• Phage Phylogeny/ supplies Phage_Concatenated_7.fasta with rooted and unrooted IQ-TREE outputs. Wolbachia Phylogeny/ contains the multilocus Wolbachia alignment and its tree.
• Pleistodontes Phylogeny/, Phylogeny Figure/ and Phylogeny Figures SI/ contain the host phylogeny, main trait-figure inputs and supplementary Newick tree files (without rendered manuscript figures), respectively. The two tree/sample identifier substitutions are listed in the Specimen-code replacements subsection below.

Sequence generation, typing and phage identification

• Hybrid Capture Baits/ contains baits-100-Relaxed-35RM-noBlastHitstoWasp.fasta, the hybrid-capture probe sequences used in the sequence-enrichment workflow. The bait-screening report PDF is not included in this deposit.
• MLST Clustering/ contains per-locus FASTA alignments and clustering scripts for gatB, coxA, hcpA, ftsZ, fbpA, wsp, cifA and cifB. For example, gatB_samples_with_partial.fasta contains partial-inclusive source sequences and gatB_clustering.R carries out locus-level grouping. The two current MLST/cif summary files are Summary new clustering_231225.csv and Summary_new_clustering_231225_with_super_profiles.csv. extend_cifA_corePlusBestPartials_auto_DEDUP_WITHIN_JRAS.R supports the partial-inclusive cifA analysis; scan_nonfunctional_cif_orthologues.py scans FASTA header labels for potentially disrupted orthologues and does not independently prove loss of function.
• PHASTEST.zip preserves per-assembly software-generated predicted prophage-region outputs, including original CSV and sequence/annotation files. The interpretation and limits of blank PHASTEST fields are described under Missing, unknown or inapplicable data and in the variable dictionary.
• wolPredictor_240526.zip contains the deposited software/example datasets, command examples, model outputs and null-model scripts. null_model_53spp.py tests prediction performance against a permuted-data reference. See the separate wolPredictor variable-conventions subsection for output naming.

Protein structures, contacts and electrostatic compatibility

• HADDOCK/ stores docking restraints and structural models, including ambig_segABC_simple.tbl, unambig_segABC_simple.tbl and HADDOCK_cifA_ub_complex.pdb. These are structural inputs/outputs rather than experimentally solved CI phenotypes.

• Protein Poses and Functional Residues.zip contains the more extensive structure-analysis workflows. Its major directories are:

  • Documentation/ — chain labels, structural models, analysis notes and provenance.
  • Primary Analysis/01_same_node_37pair_binary_analysis/ — original full-length/same-node 37-hotspot electrostatic comparisons.
  • Primary Analysis/02_partial_partner_manuscript_restricted/ — partial-partner comparisons within the manuscript's restricted analysis set.
  • Supporting Analyses/03_supporting_partial_scaffold_helpers/ — partial-scaffold and supporting 45-scaffold analyses; these are not substitutes for the primary original37 results.
  • Supporting Analyses/04_structural_electrostatic_scripts_and_models/ — structure/contact scripts and models, including the original37 hotspot provenance scripts.
  • Supporting Analyses/05_pose_interaction_metrics/ — docking/interface/contact geometry outputs and regional annotations.
  • Supporting Analyses/06_master_cif_pipeline/ — a compact higher-level pipeline and linked inputs.
  • 7_the_summary_tables/ — final presentation-oriented summary and Table 4 outputs; prefer the csv/ folder for the cleaned residue-level summary exports. The historical xlsx/ directory name can contain CSVs and does not imply that the deposited exports require Excel.

  The original scoring uses CifA = chain B, CifB = chain A in its reference binary structure, a historical CifB residue offset of 1000, and a simple charge coding (K/R/H +1, D/E −1, neutral/gap 0). Opposite-charge pairs contribute +1, same-charge pairs −1 and neutral/uncovered positions 0 to the original37 score. Coordinate-file chain conventions and residue numbering must be checked for other models. Additional model-contact and 45-scaffold work supports sensitivity/exploration; it does not recalculate or replace the manuscript's primary scoring analysis. Consult PROTEIN_MISSING_VALUE_CODING_REPORT.csv for precise NA and - conventions.

Main R workflow and external annotations

• DRYAD_R_script_standardised_bundle/ includes Analysis_Script_Pleistodontes_Wolbachia_DRYAD_standardised.R, DRYAD_INPUT_FILE_CHECKLIST.csv, and DRYAD_R_SCRIPT_MANIFEST.csv. The root-level script is an intentional duplicate. Run with the extracted data-and-code root as the working directory or use PLEISTODONTES_WOLBACHIA_ROOT, as explained in Running the standardised R analysis script. Individual subworkflows may need the appropriate R/Python packages and are not claimed to have been rerun during this curation pass.
• BV-BRC_Annotations.zip is supplied separately in the same Dryad submission. Its annotation outputs are organised under Wolbachia Output, Phage Output, Phage Genomes and Phage; the member-level directory listing is BV-BRC_annotation_contents.csv. Do not expect these large files inside the smaller data-and-code ZIP. The original .xls annotation exports are retained in their native format.

Abbreviations used in this guide

BV-BRC genome and phage annotation archive (separate file in this Dryad dataset)

The full annotation archive is supplied as a separate file in the same Dryad dataset:
BV-BRC_Annotations.zip (approximately 621 MB compressed; approximately 2.76 GB of files when expanded). It is not embedded inside this smaller data-and-code ZIP, to avoid duplicating a large compressed archive. Upload BV-BRC_Annotations.zip as a separate dataset file alongside this package.

The archive contains 1,151 files under the root directory bvbrc Annotations/. The complete internal path, filename, extension and compressed/uncompressed member sizes are listed in BV-BRC_annotation_contents.csv. The inventory is a directory listing of the archive; it does not extract or alter the annotation data.

The archive is organised into four top-level groups:

File types across the archive are: .json (469), .fasta (166), .txt (147), .gb (94), .embl (47), .genome (47), .gz (47), .gff (47), .xls (47), .html (40). The .json, .html and .txt files provide structured metadata, reports or summaries; .fasta, .genome, .gz, .gb, .embl and .gff files provide sequence, genome or annotation representations; and .xls files are tabular annotation outputs. File names and directory paths retain the identifiers supplied by the annotation workflow. Users should preserve these identifiers when matching annotation outputs to assemblies and downstream analyses.

Assemblies without corresponding BV-BRC annotation outputs. Seven assembly/sample identifiers represented in the data-and-code package have no corresponding entry in the BV-BRC annotation archive. These are samples for which Wolbachia was not detected, or for which too few sequencing reads were available to generate an assembly suitable for annotation. The absence of BV-BRC annotations for those samples is therefore expected and should not be interpreted as an accidental omission from the deposit. The original sample/assembly identifiers are retained in the data files to preserve the screening and analysis record. The manifest is a listing of outputs that were generated, not a declaration that every sampled wasp has an annotated Wolbachia genome.

The annotation outputs were generated using the BV-BRC annotation environment, including RASTtk for bacterial genome annotation; phage/viral annotation outputs were generated using the workflow described in the manuscript and Supplementary Information, including VIGOR4 where specified. Free/open tools for inspection include a web browser or text editor for .json, .html and .txt; R or Python/Biopython for FASTA and tabular processing; Artemis or UGENE for GenBank/EMBL/GFF sequence annotation inspection; and LibreOffice Calc for .xls tables. BV-BRC itself provides the original platform context for the annotation outputs.

Scope of the annotation manifest: BV-BRC_annotation_contents.csv documents the archive's complete file listing and file types, but does not contain the internal contents or column headers of the 47 .xls files. Those .xls tables are retained in their original BV-BRC-exported form in BV-BRC_Annotations.zip; their individual columns have not been independently inspected or redefined in this curation package. No content has been inferred from filenames alone.

Opening the files (free/open software)

• .csv, .tsv: text editors, R (read.csv, read.delim), Python (csv, pandas), LibreOffice Calc. Delimiters: comma for CSV and tab for TSV; UTF-8 encoding (some files have a UTF-8 BOM).
• .fasta, .afa, .fna: text editors and sequence analysis packages, e.g. AliView, MAFFT or Biopython; FASTA sequences are nucleotide or amino-acid depending on file/analysis.
• .tre, .nwk: Newick phylogenetic trees; use FigTree, iTOL, R ape or Python ETE.
• .pdb: protein 3D coordinates; use UCSF ChimeraX or PyMOL open-source edition.
• .r, .py: R and Python source; scripts may additionally need packages named within the source. The workflows were tested during the original research but were not rerun end-to-end during this curation pass.
• .json, .txt, .tbl, .bat: plain text (JSON structured outputs, program reports, HADDOCK restraints, and Windows shell commands respectively).
• .pdf, .png, .jpg, .svg: figures/plots opened with standard PDF or image readers.
• .zip: compressed nested archives; use built-in unzip, 7-Zip or unzip. To reuse source data in these archives, extract the inner ZIP before running software.

Identifiers, category keys and core variables

• JRAS... = wasp specimen/assembly identification code; not a value of infection prevalence or gene-copy count.
• W+/W- = Wolbachia detected/not detected under assay scope; noWol = no Wolbachia detected in the specified field, not proof of absolute absence.
• cifA, cifB = genes associated with CI rescue/modification; physical presence, full-length recovery, motif/structural inference and experimentally demonstrated functionality are different concepts.
• MLST = multilocus sequence typing; MLSTU = strict/unique MLST profile representation used by the analyses; cifProteins = protein-based cif grouping.
• wolB or grey map category = Wolbachia supergroup B marker as used in the geographic data, not another cif compatibility colour; consult the manuscript's Figure 6 caption.
• Blue, Yellow, Orange, Purple, Red, Green, Brown = pie-category/group indicators for Figure 6; each is a mapping code for inferred cif group membership, not necessarily a quantitative independent measurement. The Map Figure.csv rows also include Lat/Long in decimal degrees and Host abbreviation.
• Map Host codes: AUG Ficus augusta; BRA F. brachypoda; CER F. cerasicarpa; CRA F. crassipes; DES F. destruens; GLA F. glandifera; HES F. hesperidiiformes; HET F. heteromeka; LIL F. liliputiana; MAC F. macrophylla; OBL F. obliqua; PLA F. platypoda; PLE F. pleurocarpa; RUB F. rubiginosa; STE F. sterrocarpa; SUB F. subperbula; TRI F. triradiata; WAT F. watkinsiana; XYL F. xylosycia.
• Common community codes: PNG = Papua New Guinea; NTE = Northern Territory; QLD = Queensland; VAN = Vanuatu; SOL = Solomon Islands; SUL = Sulawesi; CAL = New Caledonia; WAU = Western Australia. These abbreviations identify sampling communities, not measured variables.
• Pairwise CI columns n_* are counts; prop_* and p_* are unitless proportions and p-values (0–1), respectively; column marker identifies the classification used; within means within Ficus host complex and global denotes the wider comparison set.
• For protein-structure tables: residue positions are amino-acid indices; distances are in Å unless a specific script/file says otherwise; charge scoring is dimensionless. Key conventions in the original37 pipeline: CifA is chain B, CifB is chain A, CifB original residue offset is 1000, K/R/H treated as positive, D/E negative and neutral/gap/uncovered as zero; pair_score = -(charge_cifA * charge_cifB).
• Sequence alignments use nucleotide positions or amino-acid residues depending on the analysis. Branch lengths on Newick trees use the scale in each original tree/plot, not universal calendar years.
• Protein and clustering pipeline outputs use terms such as core, partial, same_node and original37 as distinct analytical subsets; they should not be pooled without checking the scripts and inclusion criteria.

Table 4: original37 electrostatic scoring tables

Three worksheet-derived CSVs in Protein Poses and Functional Residues.zip, under Protein Poses and Functional Residues/7_the_summary_tables/xlsx/, were reformatted for machine-readable reuse. Spreadsheet title, explanatory and trailing note rows have been removed from these three CSVs only, so the first row contains the actual variable names. The complete explanatory text is preserved below. All data rows and populated scientific values were retained; subsequent missing-value coding changes to some CSVs are documented below. No scoring results were recalculated. These filenames still contain xlsx as a historical folder name, although the submitted tables are CSV.

Table_4_ORIGINAL37_original_scoring_clean_Clean_Table_4.csv — Table 4, Original37 electrostatic complementarity scores. Original analysis scoring uses same-strain cifA+cifB for full-length rows; partial rows use original cifA plus the mapped partial cifB partner.

• Full-length rows use same-strain cifA+cifB scoring from the original analysis.
• Scoring is unweighted: K/R/H = +1, D/E = −1, neutral/gap = 0; pair score = −(charge_cifA × charge_cifB). Thus opposite charge = +1, same charge = −1 and neutral/gap = 0.
• Partial rows use original cifA plus the mapped partial cifB partner sequence; N pairs is reported as covered hotspot positions / 37.
• Total score and Normalised score are unitless; N fields are counts of hotspot pairs or positions. Unavailable Normalised score for partial-partner rows means not calculated, not zero. The NA fields in the not assessed row mean that no score was assessed. For other blank pair-count fields, values were not supplied in the original output and should not be interpreted as zero.

Table_4_ORIGINAL37_original_scoring_clean_Validation.csv — Side-by-side comparison of regenerated and uploaded original-style scoring results. uploaded_ and regenerated_ fields are unitless scores or counts as their suffix states; delta_ means regenerated minus uploaded. Unavailable regenerated/delta values mean no corresponding regenerated comparison was available; do not interpret as zero.

Table_4_clean_ORIGINAL37_with_partial_sequence_scoring_Clean_Table_4.csv — Table 4, Original37 differences in matched cifA–cifB hotspot pairs relative to reference JRAS01402_0189. Original hotspot-region-first 37-pair set, with partial cifB partners scored from partial sequences.

• Original37 is the historical hotspot-region-first 37-pair set, not the NEW37 threshold-first set.
• Exact AA matches require both cifA and cifB/partial-cifB residues to match reference residues at the hotspot pair.
• Charge-compatible pairs count opposite-charge residue pairs; charge-conservative substitutions may therefore score as compatible despite amino-acid differences.
• Starred sample rows use partial cifB partner sequences; mapped/comparable hotspot positions vary, and results require more caution than full-length rows.
• Counts are numbers of residues, positions or pairs, as stated by each header; Interpretation is explanatory text. Where Hotspot pairs assessed is expressed as 34/37, it means 34 covered positions of the 37-pair hotspot set. In not assessed rows, NA means no comparison was assessed. Unavailable cifA/cifB change counts in that row are not zeros.

These three files are also indexed using their updated true headers in VARIABLE_DICTIONARY.csv and CSV_MISSINGNESS_AUDIT.csv.

Missing, unknown or inapplicable data

Intentional policy: original PHASTEST software outputs retain empty cells, rather than replacing them indiscriminately with n/a, because these CSV files may be inputs consumed by R/Python and their parsing depends on the original encoding. Empty cells represent information that was not recorded, not recovered, not assigned, not computed, or not applicable for that specific specimen/analysis. They must not automatically be interpreted as biological zero, negative detection, a confirmed absence, or missing-at-random observations. 0 is an explicit value different from a blank. Strings noWol, NA, None or other in-source codes retain their source-specific meanings and should not all be converted to missing. The variable dictionary documents the blank count for each column; where the source does not document the specific reason for a blank, its precise reason remains unknown and should be treated as missing/unspecified. No observations were invented to fill blanks.

For missing biological sequences or incomplete genes, this may arise from sequencing depth, incomplete assembly, filtering, or no recovery; it does not alone establish absence or loss of function. Some protein tables allow gaps when a partner or overlapping residue is unavailable. In the revised Summary Data Table.csv, zero values for strain-specific status indicators mean that the corresponding Wolbachia/cif strain was not detected; see the author-verified clarification below. The current table has no empty fields.

Workflow testing and reproducibility

The analytical workflows and scripts were developed, tested and used during the original research to produce the analyses and associated outputs deposited here. During preparation for Dryad, the files were organised and documented, and selected data conversions and consistency checks were verified. The complete deposited workflows were not rerun end-to-end after these curation-related changes. The scripts, input data and existing outputs are provided to support transparency, independent verification and reuse; successful execution in every user environment has not been independently established by this curation review.

Key scripts and workflow

The main R script is Analysis_Script_Pleistodontes_Wolbachia_DRYAD_standardised.R (also duplicated inside DRYAD_R_script_standardised_bundle/). The bundle's DRYAD_INPUT_FILE_CHECKLIST.csv describes expected source locations. The inspected R/Python scripts do not require .xlsx files as inputs. Two protein-analysis scripts optionally generate cifA_cifB_45_scaffold_electrostatic_groups.xlsx as an output; those output references have deliberately been retained. The deposited spreadsheet-derived tables are also available in CSV form. Analytical workflows have not been rerun end to end during curation. The deposited assembly pipeline is described in the paper/SI; Galaxy history links and raw demultiplexing shell history were not included in the lightweight archive. BV-BRC genome and phage annotation outputs are supplied in the separate BV-BRC_Annotations.zip file in this Dryad dataset. The full archive member list is provided in BV-BRC_annotation_contents.csv.

Protein and phylogeny provenance

The original 37-pair electrostatic hotspot comparison is documented in Protein Poses and Functional Residues.zip under Supporting Analyses/04_structural_electrostatic_scripts_and_models/scripts with initial_hotspot_scan_from_binary_pdb.py and hotspot_region_first_37_from_pdb.py. Separate supporting 45-scaffold analyses are additive and do not replace the original37 comparison. cifA and cifB cluster calculations and per-gene FASTA inputs are in MLST Clustering. wolPredictor_240526.zip provides a software copy, input sets, outputs, null model scripts and figures. Sequence and tree files should be linked by shared JRAS codes.

Specimen-code replacements used in tree comparisons

Running the standardised R analysis script

Run DRYAD_R_script_standardised_bundle/Analysis_Script_Pleistodontes_Wolbachia_DRYAD_standardised.R in R with the working directory set to the extracted data-and-code root (or set the PLEISTODONTES_WOLBACHIA_ROOT environment variable to that root). Consult DRYAD_R_script_standardised_bundle/DRYAD_INPUT_FILE_CHECKLIST.csv for input locations. The script uses relative project paths and checks required R packages; the deposited tables have CSV versions, while some optional Excel output references remain; no Excel input requirement was identified in the inspected scripts. The scripts have not been rerun in this curation pass. The same script is also present at the archive root for convenience.

The nested Protein Poses and Functional Residues.zip and wolPredictor_240526.zip each retain internal technical instructions and software documentation where relevant to operating their analytical subworkflows. Those files are not alternative top-level dataset READMEs.

Dataset limitations and deposited-source transparency

The BV-BRC annotation archive is included in the Dryad dataset as a separate upload (BV-BRC_Annotations.zip) because of its size; it is not embedded in the smaller data-and-code ZIP. Its 1,151 members are enumerated in BV-BRC_annotation_contents.csv. The manifest reports file names, paths, extensions and sizes, but does not expose the internal column headings of the 47 .xls annotation tables. These remain in their original BV-BRC-exported format. Nested ZIP membership and precise file paths are enumerated in FILE_INVENTORY.csv, while the annotation archive's internal paths are enumerated separately. Some scripts and figures are preserved as provenance rather than a single-click integrated pipeline. Users should cite the accepted manuscript and Dryad DOI for scientific context, and preserve specimen identifiers when repurposing the data.

Specimen-level and protein-comparison missing-value conventions

1. Summary Data Table.csv (specimen-level summary): Values of 0 in the strain-specific cif/Wolbachia summary indicators mean that Wolbachia of that cif strain was not detected. A zero is a meaningful absence code, not an empty or unknown value. The deposited CSV contains no empty fields. These absence codes should not be interpreted as proof of absence below the assay detection limit.
2. MLST Clustering/Summary new clustering_231225.csv: The deposited table excludes unused, entirely empty columns. Its retained fields are specimen_code, MLST_nogatB, and cifs; they contain no blank cells. No data-bearing columns were removed in this update, according to the author's review.
3. Protein Poses and Functional Residues.zip → Protein Poses and Functional Residues/7_the_summary_tables/csv/partial_partner_electrostatic_site_details.csv: The copy in this summary-table folder is the authoritative version. A hyphen (-) in partial_partner_changed_from_ref denotes a residue position that is a gap or was not covered in the comparison between structures. It must not be interpreted as FALSE. The same positions show partial_partner_site_covered=FALSE, partial_partner_aa=-, and gap_or_uncovered for charge/match status. A redundant edited copy from the parent summary folder is not included. The analysis-stage copy in Primary Analysis remains for provenance and also has these 245 gap/uncovered fields explicitly coded -.

These three changes affect documentation, unused blank columns and the explicit representation of inapplicable residue comparisons; they do not imply imputation of missing measurements.

Gap and uncovered positions in partial-partner comparisons

Within Protein Poses and Functional Residues.zip, both Primary Analysis/02_partial_partner_manuscript_restricted/partial_partner_electrostatic_test_results/partial_partner_electrostatic_site_details.csv and 7_the_summary_tables/csv/partial_partner_electrostatic_site_details.csv now encode a hyphen (-) in partial_partner_changed_from_ref for the same 245 of 777 hotspot comparisons. These are positions where sequence alignment introduces a gap or the partial cifB sequence does not cover the reference residue; the changed-versus-reference Boolean therefore cannot be calculated. In these records, partial_partner_site_covered=False, partial_partner_aa=-, and match_status=gap_or_uncovered. A hyphen is not False or an observed zero; the original scoring script assigns a numerical contribution of 0 to uncovered positions. Only the previously empty comparison-status cells in the analysis-stage CSV were replaced; numerical scores and all other source values are preserved.

Missing-value codes in protein-analysis tables

A missing-value audit covered Protein Poses and Functional Residues.zip. The 1,181 formerly empty cells across 26 CSV tables are represented by explicit codes. No non-empty cell or numerical result was changed during this coding. See PROTEIN_MISSING_VALUE_CODING_REPORT.csv in the root of this deposit for per-file, per-column counts and interpretations.

• - is restricted to residue comparison positions explicitly classified as gap_or_uncovered, where sequence alignment or insufficient sequence coverage prevents a residue-change assessment. It does not mean 'unchanged'.
• NA means a value was not reported, not computed, not applicable, or cannot be established from the deposited output. It is not equivalent to zero, False, or a negative biological finding. The precise reason for each NA cannot always be separated using outputs alone.
• Protein-structure metadata may omit chains, model-quality estimates or domains when they are not present or not reported. Reference-match and validation outputs can omit fields for unmatched or unassessed cases.
• Supporting 45-scaffold tables include conditional metrics that were not generated for all rows. These are explicitly coded NA; they should not be assumed to be gap/uncovered residues without corroborating alignment flags.

The original scripts were preserved, and have not been re-executed. Some scripts may regenerate empty fields upon rerunning; the deposited CSVs are accessibility-normalised copies. Software expecting blank values may need to treat NA/- as missing-value markers on import. In R, read.csv(..., na.strings=c('NA','-','')) is appropriate only where - is a missing marker rather than a literal sequence-gap character; in Python use equivalent na_values controls as appropriate.

MLST super-profile table

MLST Clustering/Summary_new_clustering_231225_with_super_profiles.csv contains 48 records, 10 named fields, and no empty cells in the author-revised version. The 12 formerly blank MLST_nogatB entries and other missing profile values use explicit noWol where appropriate. For this table, noWol indicates that a detectable/recoverable Wolbachia profile or relevant gene assignment was not available for the specified field; it is not necessarily proof that Wolbachia was completely absent. An isolate may have a cif detection while lacking a recoverable MLST profile. Column definitions and empty-cell counts are recorded in VARIABLE_DICTIONARY.csv and CSV_MISSINGNESS_AUDIT.csv. The summary file Summary new clustering_231225.csv also contains no empty cells.

Missing fields in PHASTEST outputs

The PHASTEST CSV files are retained as software-generated outputs. Blank locus_tag, PID and e-value fields mean that the source tool did not provide a value for that record; such fields may be inapplicable for some feature types (including some RNA features), or may reflect an unassigned identifier or unavailable sequence match. The exact explanation is feature-specific and has not been independently determined for every row. A blank is not equivalent to zero, a negative match, or a missing phage region. Original PHASTEST outputs have not been altered for the sole purpose of removing blanks.

wolPredictor model-output variable conventions

The variable conventions for wolPredictor_240526.zip are documented using its deposited README.md, wolPredictor_xmeans.py, wolPredictor_xmeansDelim_reduci.py, and null_model_53spp.py. Columns nSppNN_noPurge give per-taxon categorical Wolbachia predictions at a specified X-means host-clustering target before the purging step; nSppNN_PurgeTT gives predictions after indexed wolPurger threshold TT; and nSppNN_nSppKK gives the corresponding X-means host-cluster assignments, with KK the number of clusters returned. taxa is the sample/phylogeny-tip identifier, while wspClade and associated cifA/cifB/MLST fields are categorical empirical strain/profile designations. noWol in model inputs denotes no detected strain in the relevant field, not proof of absolute absence. null_53spp_metrics.csv contains one row per null-model metric (run_index) and predictive accuracy (metric, percentage). Purge suffixes index model settings rather than biological time or physical units. These are inferred from the deposited code, not independently reproduced by executing the workflow.

Additional variable definitions

VARIABLE_DICTIONARY.csv includes definitions derived from the deposited protein-analysis Python scripts and documented PHASTEST output conventions. All original scientific results, column names, missing-value codes and analysis scripts remain unchanged. Repeated definitions are supplied for each file in which a column appears.

Protein electrostatic outputs. Residue positions refer to amino-acid indices (with model-dependent PDB numbering); min_heavy_dist_A is the minimum heavy-atom distance in Å, and strength_proxy_1_over_d2 is its reciprocal square in Å⁻², a proximity proxy rather than a physical interaction energy. For the original electrostatic scoring, K/R/H are positive (+1), D/E negative (−1), and other/gap residues neutral (0). Opposite-charge pairs contribute +1, same-charge pairs −1 and neutral or uncovered pairs 0. Counts of opposite, same-charge, neutral and gap/uncovered positions are distinct. partial_partner_changed_from_ref is a residue-change comparison, with - indicating that no comparison is possible because an alignment gap or uncovered residue is present; it is not False. electrostatic_group is a qualitative classification from the relevant scoring pipeline, not a directly measured physical quantity. Chain identities inferred from chain length should not be confused with percent sequence identity.

PHASTEST annotations. start/stop and contig_start/contig_end denote nucleotide coordinates in the corresponding sequence context. strand indicates forward (+) or reverse (−) orientation; region_index identifies a predicted prophage-region group; phage_bac_class distinguishes bacterial and phage-associated calls. e-value is a sequence-match expectation value (unitless); PID is an identity-match field, but its exact scaling should be interpreted from the source export rather than assumed. GO columns refer to Gene Ontology molecular function, biological process and cellular component. Some fields are not supplied for all feature types (including some RNAs); blank values therefore must not be treated as zeros. PHASTEST's predictions are computational annotations, not experimental validation. Documentation: https://phastest.ca/instructions and https://phastest.ca/input .

Scope and limitations. Where exact thresholds or computational derivations were not established from the deposited scripts, the dictionary retains analysis-specific qualifiers rather than claiming experimentally validated semantics. The separate BV-BRC .xls annotation files remain outside this inspection because only their filename manifest was supplied.