This is a preview of subscription content, access via your institution
Relevant articles
Open Access articles citing this article.
-
Mouse phenome database: curated data repository with interactive multi-population and multi-trait analyses
Mammalian Genome Open Access 15 August 2023
-
A reaction norm perspective on reproducibility
Theory in Biosciences Open Access 25 March 2021
-
A multi-batch design to deliver robust estimates of efficacy and reduce animal use – a syngeneic tumour case study
Scientific Reports Open Access 10 April 2020
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$259.00 per year
only $21.58 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
References
de Angelis, M.H. et al. Nat. Genet. 47, 969–978 (2015).
Koscielny, G. et al. Nucleic Acids Res. 42 D1, D802–D809 (2014).
Collins, F.S. & Tabak, L.A. Nature 505, 612–613 (2014).
Crabbe, J.C., Wahlsten, D. & Dudek, B.C. Science 284, 1670–1672 (1999).
Kafkafi, N., Benjamini, Y., Sakov, A., Elmer, G.I. & Golani, I. Proc. Natl. Acad. Sci. USA 102, 4619–4624 (2005).
Richter, S.H., Garner, J.P. & Würbel, H. Nat. Methods 6, 257–261 (2009).
Benjamini, Y., Drai, D., Elmer, G., Kafkafi, N. & Golani, I. Behav. Brain Res. 125, 279–284 (2001).
Acknowledgements
This work is supported by European Research Council grants FP7/2007-2013 ERC agreement no[294519]-PSARPS (Y.B., N.K., I.G., I.J., T.S., and S.Y.) and REFINE (H.W.). We thank the International Mouse Phenotyping Consortium (IMPC) and their Data Coordination Centre for the provision of phenotyping data sets.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Integrated supplementary information
Supplementary Figure 1 The Random Lab Model.
The proposed Random Lab Model (RLM) analysis vs. the commonly-used Fixed Lab Model (FLM) standard analysis for a single laboratory experiment. The illustrated example represents a phenotyping experiment comparing two genotypes g' and g'' (e.g., a “knockout” mutant vs. its wildtype control) in a single laboratory l. The two models include the same effects (upper row), but in the RLM, the laboratory and therefore its interaction with the genotype are modeled as random (effects in red) rather than fixed (blue). bl is the contribution of laboratory specific to its measurement procedure, which is common to all animals from any genotype measured in lab l. cg'l and cg''l are the contributions of interactions of lab specifics with genotypes g' and g'' specific to measurement, which are common to all animals from same genotype measured in lab l. When phenotyping the two genotypes in the same laboratory, the laboratory effect cancels whether it is fixed or random. However, the random interaction effect are not the same, they do not cancel out, and because they are independent their variances sum up in the standard error (SE, bottom row) just as the individual animals effects do. Unlike the individual animal “noise”, it cannot be reduced by increasing the number of animals n, and it cannot be estimated in a single laboratory, and thus has to be imported from previous multi-lab experiments (GxL-adjustment). Larger SE increases the p-value and confidence interval, therefore requiring more power to show a difference, but also ensures results will be replicated in other laboratories.
Supplementary Figure 2 A proposed framework for practical community implementation of GxL-adjustment.
Researchers in a local laboratory Labl (left) perform a local phenotyping experiment comparing genotypes g' and g''. They search an online community database (right) and retrieve the current estimate of the interaction variability σ2G × L for the phenotype p of interest, estimated in other genotypes g1–g4 across other laboratories Lab1–Lab3. The researchers use this σ2G × L to GxL-adjust their local statistical analysis of p-value and confidence interval of the genotype effect, deriving a conclusion that is more likely to replicate in other laboratories. The researchers also submit their local data to the community database, thus enriching it and enabling an updated estimation of σ2G × L for future users.
Supplementary information
Supplementary Text and Figures
Supplementary Table 1, Supplementary Figures 1 and 2, Supplementary Methods, Supplementary Notes 1 and 2 (PDF 2598 kb)
Source data
Rights and permissions
About this article
Cite this article
Kafkafi, N., Golani, I., Jaljuli, I. et al. Addressing reproducibility in single-laboratory phenotyping experiments. Nat Methods 14, 462–464 (2017). https://doi.org/10.1038/nmeth.4259
Published:
Issue Date:
DOI: https://doi.org/10.1038/nmeth.4259
This article is cited by
-
Mouse phenome database: curated data repository with interactive multi-population and multi-trait analyses
Mammalian Genome (2023)
-
Increasing the statistical power of animal experiments with historical control data
Nature Neuroscience (2021)
-
A reaction norm perspective on reproducibility
Theory in Biosciences (2021)
-
Reproducibility of animal research in light of biological variation
Nature Reviews Neuroscience (2020)
-
A multi-batch design to deliver robust estimates of efficacy and reduce animal use – a syngeneic tumour case study
Scientific Reports (2020)
