sep_021.md

SEP 021 -- Experiments and Experimental Data

SEP
Title	SEP 021 -- Experiments and Experimental Data
Authors	Nicholas Roehner (nicholasroehner@gmail.com), Jake Beal (jakebeal@gmail.com)
Editor	Curtis Madsen (ckmadsen@bu.edu)
Type	Data Model
SBOL Version	2.3
Replaces	N/A
Status	Complete
Created	21-Feb-2018
Last modified	18-Sep-2018
Issue	#50

Abstract

Linking experimental data to samples and genetic designs is becoming critical to many synthetic biology projects, including the DARPA program Synergistic Discovery and Design (SD2).

Table of Contents

• 1. Rationale
• 2. Specification
  • 2.1 Experiment
    • 2.1.1 experimentalData
  • 2.2 ExperimentalData
  • 2.3 Validation Rules
• 3. Examples
  • 3.1 Interlab
• 4. Backwards Compatibility
• 5. Discussion
• 6. Relation to Other SEPs
• References
• Copyright

Rationale

• Queries for experimental results will benefit from grouping Attachment objects with classes like Experiment and ExperimentalData that are more domain-specific than Collection.
• We will eventually want a class like ExperimentalData that can be extended with indices into experimental data files and linked to experimental conditions.
• We will eventually want a class like Experiment that can be linked to an experimental design that includes experimental factors, levels, conditions, etc.

Specification

Here we propose to add two new classes, Experiment and ExperimentalData, to enable users to group links to experimental data files (Attachment objects) in a more domain-specific manner than using the SBOL Collection class.

Accordingly, we also propose to modify the relevant best-practice validation rules for specifying provenance relationships between objects in a design-build-test-learn cycle.

2.1 Experiment

The Experiment class represents a set of experimental results that were collected in accordance with some experimental design (currently not specified by this class).

Figure 1: Diagram of the Experiment class and its associated properties

2.1.1 Experiment.experimentalData

The experimentalData property is OPTIONAL and MAY contain zero or more URIs that refer to ExperimentalData objects.

2.2 ExperimentalData

The ExperimentalData class represents a set of links to experimental data files (Attachment objects) that are typically but not always associated with a single sample, experimental condition, and/or lab instrument.

Figure 2: Diagram of the ExperimentalData class and its associated properties

2.3 Modified Validation Rules for Design-Build-Test-Learn Cycles

• sbol-12506 should be changed from

"A Usage that has a roles property that contains the URI http://sbols.org/v2#test SHOULD refer to an Attachment or Collection of Attachment objects."

to

"A Usage that has a roles property that contains the URI http://sbols.org/v2#test SHOULD refer to an ExperimentalData object."

• sbol-10226 should be changed from

"A Identified with a wasGeneratedBys property that includes a reference to an Activity with a child Association that has a roles property that contains the URI http://sbols.org/v2#test SHOULD be an Attachment or a Collection of Attachment objects."

to

"An Identified with a wasGeneratedBys property that includes a reference to an Activity with a child Association that has a roles property containing the URI http://sbols.org/v2#test SHOULD be an ExperimentalData object."

Examples

3.1 Interlab Example

Figure 3 shows an example Experiment with the results of characterizing a single IGEM part. Note that the Experiment object itself is not shown, just the ExperimentalData objects to which it refers.

Figure 3: Example of an Experiment from the IGEM Interlab Study

The Experiment includes ExperimentalData derived from three Implementation objects: a positive control, a negative control, and the study construct. These ExperimentalData are generated by a single characterization Activity associated with a flow cytometer Agent and plate reader Agent. Each ExperimentalData groups the Attachment objects linking two data files: one containing absorbance values measured with the plate reader, and one containing fluorescence values measured with the flow cytometer.

Note that this example does not include sufficient detail to explicitly determine which lab instrument Agent is associated with which Attachment. To enable this determination, one could instead specify one characterization Activity per Agent and one ExperimentalData per Attachment, and then link these objects accordingly.

Backwards Compatibility

This proposal does not affect backwards compatibility.

Discussion

Relation to Other SEPs

This proposal revises the best-practice validation rules from SEP 019 that were recently incorporated into SBOL version 2.2.

References

Copyright

To the extent possible under law, SBOL developers has waived all copyright and related or neighboring rights to SEP 021. This work is published from: United States.