The reference document is available in tabular format (MIAPPE_Checklist-Data-Model-v1.1.xlsx) and PDF MIAPPE_Checklist-Data-Model-v1.1.pdf. It includes an Apendix which contain suggested values for the checklist. There are text version of the xlsx for versionning tracability (MIAPPE_Checklist-Data-Model-v1.1.*.tsv)
This document describes the MIAPPE Checklist and Data Model v1.1, a revision and extension to the MIAPPE minimal information standard published by Krajewski et al. (doi:10.1093/jxb/erv271). The revision has been published by the plant research community within ELIXIR, a pan-European federation of life science resources, in collaboration with Emphasis, the European Plant Phenotyping infrastructure. It and has four primary goals:
- The extension of MIAPPE’s scope to accommodate woody plants as an additional use-case.
- The specification of a data model for MIAPPE, to facilitate its implementation in various formats and enable its automatic validation.
- Compatibility of the MIAPPE logical standard with common frameworks for representing these data types, namely ISA-Tools and the Plant Breeding API (BrAPI).
- The enrichment of MIAPPE, with the provision of clear definitions and examples for all fields.
In this document, we review the MIAPPE data model and explain the changes motivated by the first three goals (those motivated by the last goal are self-explanatory). The overall objective is to enhance the accessibility and usability of MIAPPE, leading to its adoption by a wider community.
Figure 1 – Schematic representation of the MIAPPE data model.
The MIAPPE data model, illustrated in Figure 1, is heavily influenced by the generic ISA (Investigation-Study-Assay) data model, as interoperability with the tools from the ISA framework (https://isa-tools.org) was one of our main goals with the interoperability with the Plant Breeding API. Of its four central objects (represented in orange in the schema), three have a direct correspondence with ISA objects: Investigation and Study are homonymous in both models, while Observed Variable corresponds to the ISA concept Assay. The fourth object type, Observation Unit, is a central object in the data model of the Plant Breeding API (BrAPI), and together with Observed Variable, is important to enable MIAPPE compliant data to be served via BrAPI.
The objects Investigation and Study need further attention as they were not explicitly defined in MIAPPE 1.0, and the generic ISA definitions are broad and flexible. A section devoted to general metadata was included in the original checklist, but no formal separation between these two objects was described, in spite of references to ISA. If ISA-Tab is to be used as a format for data exchange or database submission, then Investigation and Study must be explicit objects in the MIAPPE specification, leading to their inconsistent use. The practice we propose here is that publication and submission metadata are listed at the Investigation level only, and that links to data files, timing and location data are listed at the Study level only, with the additional restriction that a Study must have a single location.
The changes can be roughly divided into four categories: terminological changes, organisational changes, scope extensions, and modelling simplifications.
These changes were aimed at making field labels less ambiguous and/or promoting interoperability with external resources. They were the following:
-
Renaming of Biosource to Biological Material which was deemed more intuitive
-
Renaming of Treatment to Experimental Factor
This matches the terminology of ISA, which uses Experimental Factor rather than Treatment to refer to this concept. The Breeding API uses Treatment to refer to the same generic concept, but each Treatment is described by a Factor and a Modality, which now correspond to MIAPPE Experimental Factor Type and Experimental Factor Value respectively. Thus, this change promotes interoperability with both ISA and the Breeding API. Furthermore, it avoids the ambiguity of the name "treatment", which may be used in a broader sense to refer to any cultivation practices. A Experimental Factor is expected to be a controlled variable, the effect of which is the object of the study.
These changes were motivated by the need to convey the data model intuitively, even implicitly in the checklist of required fields. For this to be possible, each section in the MIAPPE checklist should correspond to an object (or pair of objects, in the case of paired objects like Environment Parameter and Parameter Value) in the MIAPPE data model. Conversely, where multiple attributes of a single data object appear on the checklist, for clarity these should be listed in their own section. The changes falling under this category were the following:
-
Creation of new sections: Investigation, Study, and Observation Unit
-
Abolition of sections with no representation in the data model, and redistribution of their metadata fields into the new sections:
a. General Metadata -- fields split between Investigation and Study; some duplicated in both (e.g., Title and Description)
b. Timing and Location -- fields moved to Study section
c. Experimental Design -- fields split between Study and Observation Unit
-
Grouping unique and mandatory Environment fields under the Study, namely: Type of growth facility and Cultural practices
These changes were motivated by the need to accommodate the woody plant use-case or capture additional types of information. They were the following:
-
Addition of a License field under Investigation for FAIR compliance
-
Addition of Person section to describe parties involved in the investigation or any of its studies, including information about their roles and affiliations.
-
Addition of fields for listing Data Files and providing their version at the Study level
Data files were not listed in MIAPPE 1.0 and the version is necessary for FAIR compliance. -
Extension of the Biological Material section to accommodate woody plants:
a. Addition of geographical coordinates as a means of uniquely identifying woody plants in the field, as an alternative to accession numbers
b. Renaming of Seed Source to Material Source due to the fact that woody plants (and other types of plants also) may not be necessarily propagated by means of seeds.
c. Replacement of the fields Pretreatments and Conservation conditions, which pertained to Seed preparation, by the more generic field Biological material preprocessing which englobes both aspects. Preprocessing has been chosen instead of pretreatment to avoid the ambiguity of the name "treatment", as discussed above.
-
Addition of Event section to describe discrete occurrences that happened during the experiment, and that must be associated with a date. These can be natural or artificial interventions, can encompass elements of planning (e.g. planting) including parts of Factors (e.g. watering), or sporadic events (e.g. rain). They may apply globally, to the whole study, or locally, to specific observation units.
-
Generalisation of the description of Observation Unit.
Since the layout of experiments may vary greatly, users should be able to define their layout instead of having to provide specific levels. This has beenenabled by introducing the paired fields Spatial distribution: type and Spatial distribution: value, where users can provide specific coordinates as key-value pairs (e.g. block: 5; plot: 1; latitude: +2.341). In combination with the Observation Unit levels field under Study, this enables the description of hierarchies.
These changes were aimed at simplifying the checklist by avoiding exhaustive lists of categories for aspects such as Factors or Environment parameters. Given the broad scope of MIAPPE, the categories that make sense in one experimental setting may not make sense in another, so it is impossible to enforce mandatory fields of these types. Furthermore, it would be virtually impossible to be fully exhaustive, and thus it is best to give the users some flexibility. As such, we opted for a streamlined common representation under which any category from an exhaustive list can be provided by the users when appropriate in their particular experiments. The aspects for which this type of simplification was made were:
-
Experimental Factor, which is described using three fields: Experimental Factor type (or name), a Experimental Factor description (elaborating on the specific applications and details in free text), and Experimental Factor values (a list of all different modalities for this specific factor). The list of possible factor types is to be supplied as an appendix to the MIAPPE checklist.
-
Environment, which is described using two fields: Environment parameter name and Environment parameter value. This section supports the description of environmental conditions that were constant throughout the Study. Environment parameters that were actively measured during the experiment should be recorded as Observed variables. As in the case of Factors, a list of possible Environment parameters is to be supplied as an appendix to the MIAPPE checklist in order to guide users.
-
Samples, in which a number of fields describing sample conservation conditions (e.g. salinity, oxygenation, storage temperature) have been condensed into a single, free-text field, Sample description.