henhaobio
Sign in

bioRxivBioengineeringPreprint

Cross-species meta-analysis reveals determinants of homing gene drive performance

Sebald A. N. VerkuijliD, Edward R. Ivimey‐CookiD, B. LiuiD, Michael B. BonsalliD, Philip T. LeftwichiD, Nikolai WindbichleriD

PDFDOICodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

Homing gene drives can bias their inheritance above Mendelian expectations, but reported outcomes vary widely. We compiled a cross-species dataset of nearly one million scored progeny from 42 publications reporting CRISPR/Cas9 endonuclease-based gene drives in 10 model, pest, and vector species. Using multilevel meta-analytic models, we evaluate biological, cross, and transgene design factors as predictors of biased drive inheritance. Species is the strongest predictor, but most heterogeneity remains unexplained; design features each explain a modest fraction of the remaining variation, with large construct-to-construct differences pointing to the full combination of design choices rather than any single factor. Nuclease expression timing, the most common optimization target, has limited predictive value after accounting for correlated factors, and predictions do not transfer well between species. Maternal nuclease deposition has a marginal effect on drive inheritance but dramatically increases somatic phenotype rates in offspring, revealing a tissue-of-action rather than repair-outcome effect. An interactive web tool enables community analysis of this dataset, which will guide the design of more efficient gene drives for genetic vector control, invasive species management and other applications. https://sverkuijl.shinyapps.io/GeneDrive/

In-depth summary

Claude Sonnet 4.6 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (3 Mar 2026) · CC BY 4.0

Across nearly one million progeny from 10 species, species best predicts homing drive inheritance, while promoter, parent sex and maternal deposition each explain little of the inheritance variation.

Homing drive performance varies widely, and many design parameters change at once between implementations, so the cause of performance differences has been hard to attribute from the literature. No systematic quantitative synthesis existed; this work compiles 42 publications and fits multilevel models to separate biological, cross, and design factors.

Key results

  1. ShownSpecies is the strongest predictor of drive inheritance, with Anopheles highest and other Diptera lower, yet most heterogeneity stays unexplained.

    I^2 over 90.5%; predicted inheritance An. gambiae 95.1% (95% CI 91.6%-97.2%), An. stephensi 93.2%, Ae. aegypti 66.9% (46.5%-82.4%), Dr. melanogaster 75.8% (58.7%-87.3%); adding species ΔAIC = 21.3 (LRT p < 0.0001). · 2.2 Species is a strong, but incomplete…

  2. ShownCas9 promoter identity is a weak predictor whose rankings are largely species-specific rather than transferable between species; expression-profile classes did not clearly separate outcomes.

    Adding promoter cuts cross-to-cross variance by 10.2% (LRT p = 0.0042); species-specific promoter effects improve fit by ΔAIC = 45.6; expression class QM p = 0.1125 (n = 425); E–S–G+ 77.5% vs E+S+G+ 87.0%. · 2.3 Nuclease expression timing is poorly…

  3. SuggestedLarge between-construct variance exists but could not be attributed to target gene or insertion locus; the locus test is underpowered because most groups hold one construct.

    σ²_transgene = 0.3518, ΔAIC = 104.0; target gene explains 0% (p 0.5000); insertion locus 1% (p 0.4797), with 57% of locus groups containing a single construct; centromere proximity p = 0.7446. · 2.4 Substantial between-site heterogeneity…

  4. SuggestedDrive-parent sex and grandparental nuclease deposition have modest, species-dependent effects on inheritance; overall contrasts were not significant.

    Sex OR = 1.23 (95% CI 0.91–1.66, p = 0.1715, n = 218); Ae. aegypti female advantage OR = 1.21 (p = 0.0274, n = 71); grandparent-sex OR = 0.60 (95% CI 0.26–1.40, p = 0.2351, n = 121). · 2.5 The effect of the sex of the drive-carrying…

  5. SuggestedSomatic phenotype rates in offspring are far higher when Cas9 is inherited, with a large but non-significant maternal Cas9 effect, suggesting a tissue-of-action effect rather than altered repair outcomes.

    Cas9;gRNA vs –;gRNA F2 OR = 23.98 (p = 0.0049, n = 108); female Cas9 parent OR = 22.91 (p = 0.0675, n = 180); no F2 sex bias (OR = 0.99, p = 0.9544, n = 111). · 2.6 Somatic phenotypes reveal the reach of…

Methods, data and reproducibility

Manual extraction of cross-level progeny counts from published CRISPR/Cas9 homing drive studies, analysed with multilevel meta-analytic models on log odds of inheritance, with phylogenetic and construct random effects plus within-study paired comparisons.

  • Curated database of scored F2 progeny from published CRISPR/Cas9 homing drive crosses (2015 to mid-2025), nearly 100 metadata fields per cross · 10 species (mosquitoes, Drosophila and others) · n = 42 publications; nearly one million progeny
  • Bulk RNA-seq (embryonic, juvenile, adult) and adult germline scRNA-seq used to classify promoter gene-of-origin expression profiles · Dr. melanogaster, Ae. aegypti, An. gambiae
Methods & tools
R, metafor (escalc), orchaRd (I^2), emmeans (Tukey adjustment), Open Tree of Life phylogeny, RuleSet3 gRNA efficiency score, PubMed, Web of Science, interactive web tool
Data & accessions
https://osf.io/xcbpm
Software & environment
Analyses in R with metafor, orchaRd and emmeans; software versions listed in Supplementary Data S2 (not reproduced in main text).
Reproducibility gaps
Package versions and full model-selection results sit in supplementary files; the authors expect transcription errors may remain and plan a public change log.
Limitations (6)
  • Factors were mostly assessed one at a time, and strong correlation between research groups, species and design choices limits separating species biology from laboratory methodology. · stated by the authors
  • Inclusion required individual progeny counts from defined crosses, excluding cage trials and sequencing-only assays, so parameters such as fitness costs are not captured. · stated by the authors
  • Inheritance rate is a composite of cutting, repair pathway choice and offspring viability; the models cannot split it into constituent rates. · stated by the authors
  • Insertion-locus test is underpowered, and the phylogenetic signal sits at the parameter boundary with only 10 species, so variance partitioning is uncertain. · stated by the authors
  • Conclusions on poor transfer of predictions between species rest on model-fit comparisons; no held-out or prospective test of predictive accuracy on new constructs is reported. · noted in this summary
  • The text gives two different counts of unique records retrieved in the literature search (1,836 and 3,400), which makes the screening flow harder to audit. · noted in this summary

Read this if you design homing gene drives or model them and want quantitative evidence on what predicts inheritance bias across species.

Citations
1

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISVerkuijl '22Hammond '20Champer '20Vella '17Courtier‐Or '20Noble '18
71% semantic matchGenetics
Sebald A. N. Verkuijl, Estela González, Ming Li+6 more·Nature Communications·Nov 21, 2022· 40 citations