It is essential that any published paper includes detailed information on the following parameters to ensure that the published results are reproducible. The following section highlights the requisite details that should be reported in manuscripts using pluripotent or tissue stem cells. For a complete list of reporting recommendations, see Appendix 6.

Basic Characterization

Recommendation 5 .1 .1: Published reports should include the source of the cell line or the details of its derivation, complete descriptions of the methods used for stem cell maintenance and preservation (including culturing, passaging, freezing and thawing methods), the passage number (or ideally population doublings) of cryopreserved MCB or WCB stocks, and number of subsequent passages prior to and during experimentation.

Understanding the nature of cell materials used in experimentation is essential in the evaluation of research techniques and results, and in their comparison between laboratories. All published reports should include the unique cell line identifier, the specific source of the initial cell materials (i.e., commercial group, repository, collaborator), and detailed protocols for propagation and preservation during experimental use should be provided or referenced. Overall passage number of the cell line from derivation should be noted, and the passage duration of experimental use and its relation to characterized stocks.

Recommendation 5 .1 .2: Published reports should include the registry number of the originating cell line (hiPSC, hESC, somatic cells) and a unique number for any modification(s) made to a line, such as reprogramming to an hiPSC (see Recommendation 1 .4 .1).

Pluripotency

Recommendation 5.2 .1: Tests of pluripotency and the undifferentiated status should be thoroughly described, including assay methodology, source of reagents, readouts, and quantitation and statistical analysis, and should indicate the point in the culture history of the cell line relative to experimental studies at which assays were performed . The term ‘pluripotency marker’ should not be used to describe markers used to characterize the undifferentiated state.

Tests for pluripotency and the undifferentiated state should span the period in culture during which experiments were carried out. For example, if experiments were conducted within 10 passages of recovery of the stocks from a working cell bank, these tests should be performed early after recovery from cryopreservation and after 10 passages.

Appendix 4 provides guidelines on minimal criteria for assessment of pluripotency and undifferentiated status. However, the necessity and degree of stringency required for such an assessment will be dependent upon the context of the experiments reported and the conclusions that are drawn from them (see Figure 2, above). These are matters for assessment by reviewers, and it is important to clearly document what has been done.

Genomic Characterization

Recommendation 5 .3 .1: The specific methodology used for genotyping should be reported, including how it was performed (e .g ., number of cells analyzed) and timing (passage/population doublings) in relation to the key experiments reported.

While we do not recommend the specific methods for genotyping that should be used, the specific methodology should be described in sufficient detail so that the scope of the assays (i.e., the range of genetic variants that could have been detected) and their sensitivity (i.e., the lower limit for detecting variant cells in a mosaic culture) are clear to the reader (see Appendix 5). The author should provide a clear indication of when genotyping was carried out in relation to specific sets of experiments that generated key data for the study. In particular, the relationship between cultures of cells used for genotyping and those used for key experiments should be explicit.

Recommendation 5. 3. 2: The appearance of genetic variants during experimental procedures does not preclude publication provided that their potential effects are appropriately considered.

There is currently no general approach to predict the effect of a particular culture-acquired genomic change on traits of hPSCs or hPSC-derived differentiated cells, and on those of human somatic cells. This is because the traits may reflect the complex effects of multiple mutations, and the effects of mutations on the traits may depend on the cell type of interest and their surrounding environment. Therefore, the accumulation of knowledge on the relationship between culture-associated cellular trait changes and culture-acquired genomic mutations is important as a common resource for the stem cell research community to inform the scientifically valid interpretation of experimental results.

Stem Cell-Based Model Systems

Recommendation 5 .4 .1: Information should be reported so others can understand the work and readily compare across studies . At a minimum, this should include the source of cells or tissues, relevant disease information, if applicable, and any genetic mutations or abnormalities.

Methodological details of model systems should include enough information so that others may accurately judge the results and reproduce findings. In particular, the source of cells and/or tissue used to generate models should be provided, such as species, tissue of origin, and cell type. This includes not only donor information relevant to the model system (patient status, genotype, species) but also details of how the cells or tissues were isolated. Cells seeded from tissues should include details on biopsy site, dissociation method, a detailed characterization of starting population using recognized markers and methods, media composition, doubling rates, and phenotype of expanded cells including morphological observations and relevant molecular markers. While anonymized information relevant to the model should be included, care should be taken to avoid any potential reidentification of donors, i.e., genetic information shared through databases with restricted access.

How cells were prepared or treated before establishing the model system should also be carefully described. If cells were purified for example through FACS or other sorting strategies, details of markers used should be reported. If pluripotent stem cell lines were used as a starting point for differentiation, details of their origin should be included, such as whether they were genotyped, exhibit relevant disease mutations, and how they were cultured (media, coating material, passage number) prior to differentiation. Details of how long the cells or tissues were kept in culture (including the passaging method) before being used as a model system should be provided if known.

Recommendation 5 .4 .2: Information regarding the experimental unit, or sample type, should be reported for each experiment performed . Whether samples are individuals, cell lines, clones, tissues, organoids, batches, cells, etc., should be reported.

Quantitative analysis of model systems may include measurements across different scales. For example, when measuring a particular cellular feature (i.e., division angle) the measurement is performed on individual cells, whereas when measuring a cell population effect (i.e., number of cells of a particular identity) the measurement is performed on a tissue level. Individual experimental units (data points) should be clearly defined, whether they represent each individual cell, each organoid, each cell line, etc. For data presented as a distribution (violin plot, box plot, etc.) the individual units used to generate the data should be defined. This information will often be hierarchical for in vitro model systems (i.e., organoids derived from multiple clones, across multiple individuals) and this hierarchy should be reported.

If statistics are performed (for example, significance testing) the experimental unit should be defined (what is n?). This may include a description of technical and biological replicates, and if so, what these refer to should be explained. As a general rule, technical replicates refer to replicates of the same biological material (i.e., extracted genetic material from a single experiment) run again on the same machine, while biological replicates would represent independent biological samples. As such, technical replicates capture the variability in the assay or readout, while biological replicates reflect true biological variability.

Biological replicates may refer to many different types of replicates. For example, if protein localization was measured in 100 cells of 3 organ-on-chip models each made from 3 batches of 3 different cell lines, which of these was used for statistical comparison should be defined and this hierarchical information included, ideally even displayed in the data representation (Lord et al., 2020). The rationale for choosing the experimental unit (i.e., single organoids versus batches) should be explained. Generally, the experimental unit should be chosen based on known sources of variability in the model itself, keeping the aim of the experiment in mind. For example, if the aim is to compare a disease state to healthy control, then several different disease cell lines and several healthy control cell lines should be used to generate the model, and the experimental unit, or n, would be the number of healthy and diseased cell lines used to generate the model, rather than the model itself (i.e., the organoid or the batch).

The number of each experimental unit should be clearly defined for each experiment and statistical comparison. Ideally, pilot studies and consideration of the relevant variables should be conducted to enable adequately powered experiments (Shin et al., 2022) using appropriate power calculations (see Recommendation 4.4.3).

Recommendation 5 .4 .3: When reporting information on sample types, heterogeneity and unknown issues may arise and these should be documented as thoroughly as possible, including both known sources of heterogeneity and unknown, in which case the lack of relevant information that may influence heterogeneity should be reported.

Depending on the model system, the source of cells or tissue may contribute to phenotypic heterogeneity. This is especially relevant to tissue stem cell-derived model systems, such as tissue stem cell-derived organoids. In particular, samples taken from different parts of the same tissue may exhibit heterogeneous phenotypes. Tumor samples and other tissues with genetic abnormalities may especially exhibit such heterogeneity and so whether these sources of heterogeneity are present, and details of where cells originate from within the tissue, should be provided. However, this information may not be present, so for patient samples with missing information, those unknowns should also be outlined. Thus, if the source of heterogeneity is known, for example related to tissue of origin (i.e., tumor samples, genetic abnormality, etc.) this should be reported. If heterogeneity is likely present but relevant information is missing, that lack of information should be stated.

Recommendation 5 .4 .4: Publication of phenotypes that use computationally derived classifiers should include the data and annotated code used for phenotype classification. Researchers must follow FAIR (https://www .go-fair .org/fair-principles/) and CARE (https://www .gida-global .org/care) data management principles. Computational phenotyping can be very effective for benchmarking cell models, or discriminate confounding generic cell responses from cell or tissue-specific effects of an assay.

Computational phenotypes are derived from analysis of high-dimensional molecular data, which may include a combination of image-analysis, antibody panels, (phospho)proteomics, metabolomics, transcriptomics, or epigenome profiling. While several computational approaches may be gainfully employed to benchmark or evaluate cellular responses, all require reporting of experimental variables using FAIR data principles. This reporting should include the use of structured ontologies of metadata, cell line accession numbers, and standardized data formats that allow for parsing of experimental series into standardized computational workflows. Best practice includes using markdown to explain key steps in the code and publication of workflows in code notebooks or git-libraries. Documentation of phenotyping markers derived from these workflows should include evaluation of marker sensitivity and specificity when applied to independent datasets and marker recall when code is applied to new data.