Data from: Anatomy and mechanics of tsetse fly blood feeding
Data files
Sep 22, 2026 version files 89.46 MB
-
README.md
11.40 KB
-
tsetse_quantitative_data.zip
89.45 MB
Abstract
African trypanosomiasis is a neglected tropical disease transmitted by tsetse flies (Glossina spp.) in sub-Saharan Africa. The fly’s saliva carries parasitic unicellular trypanosomes, such as Trypanosoma brucei, leading to infections in humans, domestic animals, and wildlife. Tsetse flies feed on a broad spectrum of hosts, including humans, elephants, buffaloes, rhinos, hippos, turtles, and monitor lizards. To understand how the insects can penetrate such diverse skin types, we detailed the anatomical structures involved in tsetse blood feeding, their mechanical properties and the forces exerted by the fly during probing. We found that the tsetse fly’s feeding apparatus does not rely on exceptionally unique structures or extraordinary forces. Instead, the tsetse fly has evolved subtle yet highly efficient adaptations, which include a labellum equipped with arrays of small teeth. When combined with a robust retraction movement of the proboscis, these structures create lesions on various types of skin, thus facilitating blood pool feeding on a broad host spectrum.
Dataset DOI: 10.5061/dryad.n02v6wxdh
Uploaded files
- tsetse_quantitative_data.zip: archive containing the original supplementary workbooks, force recordings, CSV exports, variable dictionaries, documented reconstructions and known limitations. Its top-level folder is
tsetse_quantitative_data/. The folders described below are inside this archive. The archive also contains its own README and file checksums. - README.md: this separately uploaded overview of the dataset and its archive contents.
Scope and version
Documentation revised 22 September 2026 during Dryad curation. The original data and previously supplied reconstructions are unchanged. This archive covers the available quantitative attachment, proboscis-force, and blood-meal data. It preserves the three supplementary workbooks and the supplied force-acquisition archive, and provides open-format access copies and explicitly documented reconstructions. It does not claim to contain the complete microscopy, video, or anatomical reconstruction data for the paper.
Related article: https://elifesciences.org/reviewed-preprints/110151v1. Figure labels follow the supplied revised manuscript and final submission artwork: Figure 2, Figure 8, Figure S4, and Figure S5. Dataset creators, affiliations, contact, funding, and final related-publication identifiers belong in the repository metadata and are not inferred from file ownership.
Files and use
quantitative_data/: three unchanged original supplementary XLSX files, eight saved-value CSV sheet grids, variable dictionary, workbook inventory, original-to-renamed file manifest and figure mapping. Read its README for table layout and units.additional_force_data/originals/: 249 unchanged AcqKnowledge recordings/derived spectral files, 11 source XLSX files and two Prism projects. Original relative filenames are retained.additional_force_data/csv/: 249 channel exports. The 227 mN channels have elapsed time and recorded force; 22 dbV spectral channels have sample index and value. File counts are not independent-fly counts. See its README andchannel_dictionary.csvfor acquisition units and limitations.additional_force_data/source_manifest.csv: source-to-package paths and hashes.force_values_crosswalk.csvlinks all 400 supplementary probing-force observations to source cells.figure8a_source_candidates.csvgives exploratory matches for the illustrated traces, not definitive processing instructions.documented_reconstructions/: separately labelled access copies, source-cell correction logs and recomputation tables. These do not overwrite the supplied observations. ReadKNOWN_LIMITATIONS.mdbefore using them.KNOWN_LIMITATIONS.md: evidence for the reconstructions and unresolved interpretation limits.SHA256SUMS.txt: SHA-256 of each archive member except this checksum file.
Original workbooks are authoritative records of what was supplied, including errors; their preservation does not certify correctness. CSV sheet grids retain Excel row/column positions, headers and summaries. They are not tidy observation tables. Empty cells are not zeros. Formula values in original CSV exports are saved cached values without recalculation; formulas and formatting remain available in XLSX.
Quantitative coverage
| Data | Observations | Units | Figure |
|---|---|---|---|
| Attachment | 558 listed trials; 20 flies, 10 per sex; seven surfaces; 140 fly–surface groups | rpm, cm, mg, mN; dimensionless safety factor | 2 |
| Penetration and retraction | 200 values each, nine substrates, 20–25 values per substrate | mN | 8C–F; S4 |
| Blood meal | 24 feeding events | mg/s, mg, s; original uptake-rate header conflicts with mass-rate interpretation | S5 |
Repeated trials and observations across surfaces are not independent flies. Identifiers are local to each source unless a crosswalk establishes otherwise. Do not concatenate raw recording blocks, compiled tables and summaries as separate observations.
Reconstructions and uncertainty
The source-supported access copy restores 15 shifted attachment input fields, without changing any force or safety-factor values. Omitting the inconsistent fourth Male3/3-µm trial reproduces all four stored summary statistics for that group and all 280 stored attachment means match within rounding. The historical reason for that omission remains unknown. Five stored standard deviations differ from reconstruction.
For Fly16, a dimensionless source formula was transferred as a drinking rate. The separate access copy restores mass divided by duration and its rate summaries; it explicitly uses mg/s. Fly8 apparently contains a start timestamp instead of drinking duration, but the 17.28-s source interval differs by 0.005 s from the interval implied by its supplementary mass and rate. Fly8 is therefore not replaced in the access copy. Fly6 and smaller Fly13/Fly24 inconsistencies remain unchanged and flagged. No missing values were imputed and no observation was deleted from the preserved records.
These limitations affect reuse and require transparent reporting. In particular, the supplied final Figure S5 does not incorporate the documented Fly16 restoration. This package contains data and documentation; it does not assert that manuscript figures have already been revised. The deposit is not a complete reanalysis or validation of every statistical annotation.
Software and access
XLSX files can be opened in spreadsheet software; CSV uses UTF-8, comma delimiters and decimal points. CSV.GZ uses gzip. AcqKnowledge files preserve proprietary metadata and markers that are not all represented in exports. ACQ decoding used bioread 2025.5.2 (https://github.com/uwmadison-chm/bioread). Channel values were exported with 17 significant digits and read back for exact comparison with decoded arrays. This checks conversion fidelity, not sensor calibration. The original methods name Throw the fly 1.0, AcqKnowledge/BIOPAC, Excel 2021 and GraphPad Prism 9.0.
Reuse terms
The data in this Dryad deposit are provided under the Creative Commons CC0 1.0 Universal Public Domain Dedication: https://creativecommons.org/publicdomain/zero/1.0/. This README imposes no additional restrictions on reuse. Proprietary source-file formats do not impose an additional reuse condition on the deposited data. Referenced external software remains subject to its own licenses and is not included or relicensed by the deposit.
Detailed file documentation
The exact uploaded archive is tsetse_quantitative_data.zip. After extraction, begin in its tsetse_quantitative_data/ folder. The separate README.md is identical to the archive-root README.
file_catalog.csvlists and describes every file in the archive, with its exact path, related source/access copy and software/access route. This includes each of the 249 ACQ files and each of their 249 CSV exports individually.additional_force_data/README.mdexplains the recording, calculation and summary workflow and how to navigate the force collection.additional_force_data/documentation/FILE_NAMING.mddefines folder names, substrate aliases, identifier conventions and the limits of suffix interpretation.additional_force_data/documentation/WORKBOOK_GUIDE.mddescribes all 11 source XLSX workbooks, irregular calculation blocks and both Prism projects.additional_force_data/documentation/workbook_sheets.csvinventories all 174 source worksheets.workbook_cells.csv.gzexposes their 16,051 nonempty source cells, including formulas and cached results.additional_force_data/documentation/workbook_label_dictionary.csvlists all 776 distinct nonempty labels/annotations found in those workbooks, with meanings, units or categorical interpretations and exact source locations. It explicitly separates defined variables, local IDs, notes and quantities whose interpretation remains uncertain.additional_force_data/documentation/prism_tables.csvlists all ten embedded Prism tables;prism_table_values.csvprovides the 2,823 XML Y-value slots, including blanks. Slots and repeated tables are not counts of independent observations.additional_force_data/documentation/TABLE_SCHEMAS.mddefines every field in the force channel exports, metadata tables and new open-format registers.additional_force_data/examples/read_acq_example.pyis a working example for reading the originals using bioread without BIOPAC. Its accompanying README supplies installation commands, a runnable example and standard-library CSV access. Existing CSV exports need no proprietary software.
The original three supplementary workbooks and their variables remain documented in quantitative_data/README.md, variable_dictionary.csv, workbook_inventory.csv, file_manifest.csv and figure_data_map.csv. The separately labelled reconstructions are documented in documented_reconstructions/README.md and KNOWN_LIMITATIONS.md.
Missing values and uncertain interpretation
Blank cells are unpopulated/missing, not zeros. Source strings n.a, n.a. and na likewise mean a value was not supplied; their exact missingness reason is not encoded. CSV grid exports include headers and summary rows, not just observations. ACQ/CSV file counts, worksheet counts and Prism table slots must not be interpreted as independent sample sizes. Avoid combining original measurements, their exports and repeated summary tables as if they were different observations.
Historical sources contain unlabelled working cells, subjective event notes and incomplete spectral settings. Their original positions, contents and formulas are retained; the documentation identifies uncertainty instead of assigning invented physical meanings. For reproducible force comparisons, use the supplementary force table and supplied source-cell crosswalk; for reanalysis of recordings use the raw force channels and separately document any new interval selection or baseline correction.
Documentation revision
This revision clarifies CC0 reuse, adds complete file navigation and source-variable documentation, supplies open cell/table registers and a tested Python access example. It does not change any pre-existing source file, numerical CSV export, reconstruction or scientific result. Public release remains subject to Dryad curation. Guidance used: https://datadryad.org/docs/README.md and https://datadryad.org/help/guides/best_practices.
