Add a pid property to SampleType and change the uniqueness constraint

[RKrahl]

The following proposal for a minor schema change is a result of a discussion with Kirsty Syder, Steve Collins and myself. Credits should go to Steve for bringing it up.

The molecularFormula property in SampleType is not nullable, because it is part of the uniqueness constraint. This is unfortunate because for some samples, maybe even for most, molecularFormula is meaningless. Of course we can work around this by setting something like N/A or :unap to denote "not applicable", but that is probably only the second best solution.

Ideally, the sample type should use a controlled vocabulary, such as an ontology. In this case, we'd want to link any SampleType object with the corresponding term in that vocabulary. Hence, we'd need to add a pid property to do that. Consequently, we should use that pid property also in the uniqueness constraint, which would allow to change molecularFormula to be nullable.

As an interim solution in the absence of any controlled vocabulary for sample types, we could populate the pid attribute with some arbitrary local id that only needs to be unique and is otherwise an opaque string.

[spc93]

Sounds good! It might be some time before we have PIDs for all salple types.

[RKrahl]

Note that this is partly a duplicate of #239.

[RKrahl]

In order to provide the details, the proposal is to change SampleType as follows:

SampleType

A type of sample

Constraint: pid

Relationships:

Card	Class	Field
0,*	Sample	samples

Other fields:

Field	Type	Description
name	String[255] NOT NULL
pid	String[255] NOT NULL	A persistent identifier attributed to this sample type, ideally referring to a vocabulary term
description	String[255]	An informal description of the sample type
molecularFormula	String[1023]	The formula, if applicaple, either as free text string or International Chemical Identifier (InChI) or other standardized form
safetyInformation	String [4000]	Any safety information related to this type of sample

[RKrahl]

As discussed in the ICAT Schema Discussion on April 2nd and in the collaboration meeting today, this change should be in a version 7.0 release that we aim to make in the second half of this year.