INFRAFRONTIER quality principles in systemic phenotyping

Improving reproducibility and replicability in preclinical research is a widely discussed and pertinent topic, especially regarding ethical responsibility in animal research. INFRAFRONTIER, the European Research Infrastructure for the generation, phenotyping, archiving, and distribution of model mammalian genomes, is addressing this issue by developing internal quality principles for its different service areas, that provides a quality framework for its operational activities. This article introduces the INFRAFRONTIER Quality Principles in Systemic Phenotyping of genetically altered mouse models. A total of 11 key principles are included, ranging from general requirements for compliance with guidelines on animal testing, to the need for well-trained personnel and more specific standards such as the exchange of reference lines. Recently established requirements such as the provision of FAIR (Findable, Accessible, Interoperable, Reusable) data are also addressed. For each quality principle, we have outlined the specific context, requirements, further recommendations, and key references. Supplementary Information The online version contains supplementary material available at 10.1007/s00335-021-09892-2.


•
We apply the DIRECTIVE 2010/63/EU including its national transpositions (or as applicable the respective Canadian and UK legislations) and other relevant Animal Health and Welfare policies Requirements: • Site specific transposition and application of relevant international and national regulations and requirements (instructions etc.) • Ethical review process by competent authorities, both internal and external ○ acceptability of research projects ○ follow progress of research projects ○ perform a retrospective review ○ review animal care and accommodation standards ○ evaluate the "cost/benefit" balance of research projects ○ verify compliance of research projects with legal requirements for replacement, reduction, and refinement

INFRAFRONTIER Quality Principle (2) in Systemic Phenotyping
(2) We promote and apply the 3Rs (Replacement, Reduction, Refinement) Context: • Characterisation of genetically modified rodent models for human diseases has become a key tool in basic and biomedical research to understand molecular mechanisms of human disorders and for the development of new therapies • The laboratory mouse shares over 90% of its genome with humans, as well as similar physiology and anatomy for the majority of organs and organ systems • A priori power analysis (minimum number of animals required) • Proper age-matched controls on the same genetic background (ideally littermates) • Justify the genetic background: appropriate or not for a given test?
• Be aware about the health status and potential interference with the phenotyping assays

INFRAFRONTIER Quality Principle (4) in Systemic Phenotyping (4) We apply Standard Operating Procedures
Context: • Key principle: to reach repeatability, reproducibility, and robustness at each step of the experiment (e.g. performing the assay, collecting data, analysing data), ensure transparency as well as management and transfer of knowledge Requirements: • Implement and maintain SOPs for phenotyping tests and quality-related activities (e.g. data management, data analysis, animal welfare, training of staff) • Define ○ responsibilities, operational methods, parameters, and metadata to be measured, recorded, evaluated, and reported, including quality control steps ○ controlled equipment and resources / consumables • Periodic SOP review and SOP management (disseminate/train users on the latest SOP version, withdraw outdated versions)

Recommendations:
• Standardisation of procedures or phenotyping pipelines where appropriate (e.g. IMPC) • Evaluate repeatability and reproducibility of assays (e.g. use one versus several batches of mice; inter-operator reproducibility) • Report any deviations and assess them as appropriate

6/15
References: Recommendations: • Training activities should be regularly and independently reviewed / audited • As far as possible, provide more than one person with the competencies to carry out a given procedure to limit operator effect • The reference range reflects the biological / technical variation of a parameter by analysing control groups and is specific to a laboratory and its procedures • Reference ranges are important tools to ensure quality of data (monitoring over time, identification of technical problems, and biologically impossible outliers) and to identify systematic shifts • A reference range is possible only with a large-scale project (e.g. n >40, depending on test) • Reference ranges may also be used to identify potential phenotypes in experimental animals (see principle 8, reference range model)

Requirements:
• Reference ranges must be established on control mice using the same procedures, conditions, age, sex, and genetic background • Control (if possible littermates) and experimental mice should be tested concurrently

Recommendations:
• Record and analyse the variability of a given parameter • Apply the same procedure at the same time, age, sex, background (common pipeline between centres) • Collect and consider differences in the metadata (e.g. housing conditions, diet formulation and batch number) • Analyse the inter-centre heterogeneity visually and by using meta-analytical measures (see reference) • Perform regular re-evaluations, e.g. by analysis of current data, on a periodic basis (e.g. 5 years), and decide on the necessity to re-analyse the reference line • Statistical power analysis a priori to determine the minimum cohort size required for the tests (consider sexual dimorphism or other co-variants like body weight)

Recommendations:
• Exchange the same mutant line or biological samples between centres, especially when new tests have been implemented • When discordance between centres occurs, consider metadata differences and the possibility of false negative or positive results

(8) We use appropriate statistical analyses that are fit for purpose
Context: • Statistical analysis is the process of generating statistics from stored data and analysing the results to deduce or infer meaning about the underlying dataset or the reality that it attempts to describe. In addition, it empowers to ensure reproducibility while minimising the number of animals required.
• Statistical analysis may be used to: • Statistical power analysis a priori to determine the minimum cohort size required for the tests and a posteriori to validate the hypothesis • Statistical analysis should be performed and reported (e.g. test method, software used)

9/15
• Analyse based on the assay and the structure of the data • Know the assumption and conditions of the statistical methods

•
Know the type of the data collected and objective of the study

Recommendations:
• Work with biostatisticians to define the most appropriate analysis method • Use current version of PhenStat tool (see Kurbatova et al 2015Kurbatova et al , 2020 • Apply soft windowing for large datasets with running baseline (see Karp et al 2014) References: • ARRIVE guidelines (

10/15
Requirements: • Record the health status, housing and husbandry conditions and capture them for data analysis (e.g. diet, bedding, day-night cycle, air renewal) • Record the operational conditions and capture them for data analysis (e.g. number of mice per cage, mutants and controls grouped or not?) • Record the experimental conditions (e.g. fasting time and duration, which anaesthesia, route of blood collection, time of analysis) • Record equipment information to be tracked (e.g. ID, Manufacturer, Model) Context: • PDCA (Plan-Do-Check-Act) is an iterative, four-stage approach for continually improving processes, products, or services, and for resolving problems. It involves systematically testing possible solutions, assessing the results, and implementing the ones proven to work.
The PDCA Cycle provides a simple and effective approach for solving problems and managing change. It enables businesses to develop hypotheses about what needs to change, to test these hypotheses in a continuous feedback loop, and to gain valuable learning and knowledge.
The PDCA cycle consists of four components: • INFRAFRONTIER: to improve our processes we perform inter-centre exchange of experience on quality management and principles through working groups, also considering novel approaches and technologies as discussed at INFRAFRONTIER/IMPC phenotyping workshops and scientific meetings

Requirements:
• Commitment to quality policy and principles from all personnel at all levels • Implement quality related processes that enable the consistency and effectiveness of the quality management to be checked against the set objectives, e.g. Assign an independent internal reviewer to assess processes and recommend improvements o Invest in training of quality awareness (also refer to Quality Principle (5)) • The outcome of quality related processes are corrective and/or preventive actions and action plans to support continuous improvement