Skip to main content

Table 1 The support for different tasks offered by different modelling structures: NVT (Name-Value-Type), ontologies, external files and extend model inheritance (EMI).

From: An analysis of extensible modelling for functional genomics data

Extension

Support for task: Search

NVT

Different sources will differ in attribute and value, therefore good for local data because NVT can be used to encode arbitrary properties as long as local users are aware of the data types that can be searched. Poor for non-local searches, as inconsistent attributes and values are likely to be used.

Ontology

Okay if searched with exact matching terms; more difficult to support non-exact match because the search engine is unlikely to search within the ontology structure.

External file

Not good; there may be no access to the structure of the file. Only information retrieval style requests can be made.

EMI

Extensions can be searched locally but non-local searches will not be possible unless the extended models are shared.

Extension

Support for task: Share

NVT

Good for local sharing, poor for sharing externally because properties may be encoded in NVT in inconsistent ways.

Ontology

Good if terms agree (if the same ontology has been used).

External file

Okay if file is in a standard format, otherwise bad (information may be difficult to access).

EMI

Good for local sharing; cannot be shared externally unless the extended models are shared.

Extension

Support for task: Read

NVT

Generally good because writer can be expressive (NVT is better than plain text); only problem is misinterpretation if NVT is used inconsistently.

Ontology

Good because terms are well defined.

External file

Good if file is in a standard format, otherwise bad. Other software may be required to access the file, such as for images, archive files, spreadsheets and so on.

EMI

Good because writer can be as expressive as required.

Extension

Support for task: Repeat Experiment

NVT

Okay for local case (especially good if data capture is automated); in general it is a hard problem for the non-local case.

Ontology

Good for the non-local case. May be less good for local case if local terms are converted to ontology terms and cannot be converted back (ontology may not be able to express all local data in a lossless manner).

External file

Okay if file is in a standard format, otherwise bad.

EMI

Good for local case, poor for non-local case unless extensions are widely shared.

Extension

Support for task: Compare experiments manually

NVT

Okay, but inconsistencies could be problematic if data types are encoded differently in different settings.

Ontology

Good because terms are well defined and standard.

External file

Okay if the file is in a standard format that can be easily processed.

EMI

Generally good because the model developer can be expressive.

Extension

Support for task: Compare experiments automatically

NVT

Good for local case; not good for the non-local case because NVT is likely to have been implemented differently.

Ontology

Good (consistent representation from different experiments).

External file

Okay if data are stored in a spreadsheet or tab-delimited text and descriptive metadata are stored correctly within the data format, or if the external file is in a standard format that can be easily processed.

EMI

Good for local case; cannot be done for the non-local case unless the extensions are widely shared.

Extension

Support for task: Query

NVT

Worse than problem for search because queries are generally more precise.

Ontology

Generally good, but must query more than one language and the software for query evaluation may not be able to call out to a reasoning service (to make use of the ontology structure).

External file

Not good (it must be assumed that there is no access to structure).

EMI

Good for local case, cannot be queried non-locally unless the extensions are shared.

Extension

Support for task: Analyse

NVT

Not possible; generic analyses must not depend on such data.

Ontology

May not be relevant; analysis is not usually over ontology terms (but much better than NVT if it is).

External file

Okay if data are stored in a spreadsheet or tab-delimited text and metadata are stored correctly within the data format, or if the external file is in a standard format that can be easily processed.

EMI

Okay for local analysis but additional wrappers may be required to allow generic analysis software to access the data. Poor for non-local case as the format will have to be interpreted and software must be written.

Extension

Support for task: Browse

NVT

Okay (probably better than plain text).

Ontology

Good, less chance of misinterpretation than NVT.

External file

Not good unless file is immediately readable.

EMI

Good because writer can be expressive.

Extension

Support for task: Populate

NVT

Easy to populate but hard to enforce consistency.

Ontology

Easy as long as ontology is in place and easily accessible.

External file

Easy to populate but hard to enforce consistency.

EMI

Easy.